{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import os\n", "import sys\n", "\n", "sys.path.append(\"..\")\n", "\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", "import pandas as pd\n", "import seaborn as sn\n", "import utils\n", "from sklearn.preprocessing import StandardScaler\n", "\n", "plt.style.use(\"ggplot\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Tiltaksovervakingen: opsjon for kvalitetskontroll av analysedata\n", "## Notebook 3: Outlier detection for whole water samples\n", "\n", "Exploring distributions for **single parameters** (as in notebook 2) is a reasonable starting point for quality assurance, but a more general approach is to look for \"outliers\" at the **water sample** level i.e. samples that are of questionable quality because *one or more* parameter values are unusual. If the suite of water quality parameters analysed is consistent (i.e. no data gaps), then each sample can be considered as a point in $n$ dimensional space, where $n$ is the number of parameters measured. Rather than looking for \"outliers\" along a single dimension (as in the distribution plots considered already), we can instead look for \"outliers\" in this higher dimensional space. A variety of algorithms are available to do this, some of which are explored below. " ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 1. Terminology: \"outlier\" versus \"novelty\" detection\n", "\n", "\"**Outlier**\" and \"**novelty**\" detection are two different kinds of **anomaly** detection i.e. where we are interested in detecting abnormal or unusual observations. The difference between the two is important, but often overlooked:\n", "\n", " * **Outlier detection:** We have a **single dataset** that is believed to contain \"outliers\", which are observations that are \"far\" from the others (for some chosen definition of \"far\"). Outlier detection estimators thus try to fit the regions where the data is most concentrated, ignoring the deviant observations.\n", "\n", " * **Novelty detection:** We have access to a **reference dataset** that is *not* polluted by outliers, and we are interested in detecting whether a new observation (from a second dataset) is an outlier - a \"novelty\" - or not\n", " \n", "In the context of this project, we are primarily interested in **novelty detection**, because we have a reference historic dataset extracted from Vannmiljø and we would like to gauge whether observations in the \"new\" dataset are sufficiently unusual/unlikely to warrant further investigation and reanalysis. However, the [summary in the previous notebook](https://nbviewer.jupyter.org/github/NIVANorge/tiltaksovervakingen/blob/master/notebooks/02_distribution_plots.ipynb#3.-Summary) identified some possible issues in the Vannmiljø data, as well as in the \"new\" data. I am therefore reluctant to use the historic Vannmiljø dataset as a \"reference\" without additional cleaning. Instead, to begin with, at least, I will combine the \"new\" and \"historic\" results into a single dataset and then perform **outlier detection** (not novelty detection). This can be revised later if desired." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 2. Read data" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "# Choose dataset to process\n", "lab = \"Eurofins\"\n", "year = 2022\n", "qtr = 3\n", "version = 1" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "fold_path = f\"../../output/{lab.lower()}_{year}_q{qtr}_v{version}\"" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
vannmiljo_codesample_datelabperioddepth1depth2ALK_mmol/lANC_µekv/lCA_mg/lCL_mg/l...N-NO3_µg/l NN-TOT_µg/l NNA_mg/lP-TOT_µg/l PPH_<ubenevnt>RAL_µg/l AlSIO2_µg/l SiSO4_mg/lTEMP_°CTOC_mg/l C
0002-1059612022-07-06Eurofinsnew0.00.00.0395.02.11.5...5.0260.00.4717.05.567.0701.3149130.40NaN19.0
1002-1059612022-07-19Eurofinsnew0.00.0NaNNaN2.5NaN...NaNNaNNaNNaN5.9NaNNaNNaN15.0NaN
2002-1059612022-08-03Eurofinsnew0.00.00.06130.02.71.4...5.0250.00.3517.06.248.0607.8062580.2715.016.0
3002-1059612022-08-12Eurofinsnew0.00.00.04110.02.31.5...5.0330.00.4817.05.661.0654.5605860.32NaN21.0
4002-1059612022-08-30Eurofinsnew0.00.0NaNNaN2.6NaN...NaNNaNNaNNaN5.7NaNNaNNaN13.0NaN
\n", "

5 rows × 25 columns

\n", "
" ], "text/plain": [ " vannmiljo_code sample_date lab period depth1 depth2 ALK_mmol/l \\\n", "0 002-105961 2022-07-06 Eurofins new 0.0 0.0 0.03 \n", "1 002-105961 2022-07-19 Eurofins new 0.0 0.0 NaN \n", "2 002-105961 2022-08-03 Eurofins new 0.0 0.0 0.06 \n", "3 002-105961 2022-08-12 Eurofins new 0.0 0.0 0.04 \n", "4 002-105961 2022-08-30 Eurofins new 0.0 0.0 NaN \n", "\n", " ANC_µekv/l CA_mg/l CL_mg/l ... N-NO3_µg/l N N-TOT_µg/l N NA_mg/l \\\n", "0 95.0 2.1 1.5 ... 5.0 260.0 0.47 \n", "1 NaN 2.5 NaN ... NaN NaN NaN \n", "2 130.0 2.7 1.4 ... 5.0 250.0 0.35 \n", "3 110.0 2.3 1.5 ... 5.0 330.0 0.48 \n", "4 NaN 2.6 NaN ... NaN NaN NaN \n", "\n", " P-TOT_µg/l P PH_ RAL_µg/l Al SIO2_µg/l Si SO4_mg/l TEMP_°C \\\n", "0 17.0 5.5 67.0 701.314913 0.40 NaN \n", "1 NaN 5.9 NaN NaN NaN 15.0 \n", "2 17.0 6.2 48.0 607.806258 0.27 15.0 \n", "3 17.0 5.6 61.0 654.560586 0.32 NaN \n", "4 NaN 5.7 NaN NaN NaN 13.0 \n", "\n", " TOC_mg/l C \n", "0 19.0 \n", "1 NaN \n", "2 16.0 \n", "3 21.0 \n", "4 NaN \n", "\n", "[5 rows x 25 columns]" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Read from SQLite\n", "stn_df, df = utils.read_data_from_sqlite(lab, year, qtr, version)\n", "\n", "# # Subset data to just the quarter of interest\n", "# months_dict = {\n", "# \"q1\": [1, 2, 3],\n", "# \"q2\": [4, 5, 6],\n", "# \"q3\": [7, 8, 9],\n", "# \"q4\": [10, 11, 12],\n", "# }\n", "# months = months_dict[qtr]\n", "# df = df[df[\"sample_date\"].dt.month.isin(months)]\n", "\n", "df.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 3. Select parameters\n", "\n", "In order to perform outlier detection in a multi-dimensional space, it is necessary that all water samples have a **complete set of values for all parameters**. This is because outlier detection algorithms work by calculating distance metrics between samples, and this is not possible if a sample can't be located along one or more of the dimension axes due to missing values. It is therefore necessary to choose a set of parameters where the data are complete.\n", "\n", "The code below calculates the percentage of the time that each lab measures each parameter, where the percentage is calculated as:\n", "\n", " 100 * number_of_samples_for_par_X_from_lab_Y / total_number_of_samples_from_lab_Y" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
ANC_µekv/lCA_mg/lCL_mg/lILAL_µg/l AlKOND_mS/mK_mg/lLAL_µg/l AlMG_mg/lN-NO3_µg/l NN-TOT_µg/l NNA_mg/lP-TOT_µg/l PPH_<ubenevnt>RAL_µg/l AlSIO2_µg/l SiSO4_mg/lTEMP_°CTOC_mg/l C
lab
Eurofins29.250457100.00000029.25045757.67824557.67824529.25045752.37660029.25045729.25045729.25045729.25045729.25045799.90859257.67824529.25045729.25045720.56672857.678245
Eurofins (historic)27.47778573.95762126.93096456.32262556.66438834.3814080.00000025.56391027.13602225.08544125.15379425.01708893.43814156.5276830.00000027.54613827.61449156.459330
NIVA (historic)15.17643597.58522715.75956936.32625697.54784715.7520930.05233315.77452215.73714115.77452215.77452215.75956997.74222536.32625611.81967715.7595695.21082515.774522
VestfoldLAB (historic)17.49632199.87314217.82107928.81209717.81600417.81600428.35540717.8160043.00400917.46080117.81600416.20743999.87314236.7635870.00000017.81093018.17628322.702593
\n", "
" ], "text/plain": [ " ANC_µekv/l CA_mg/l CL_mg/l ILAL_µg/l Al \\\n", "lab \n", "Eurofins 29.250457 100.000000 29.250457 57.678245 \n", "Eurofins (historic) 27.477785 73.957621 26.930964 56.322625 \n", "NIVA (historic) 15.176435 97.585227 15.759569 36.326256 \n", "VestfoldLAB (historic) 17.496321 99.873142 17.821079 28.812097 \n", "\n", " KOND_mS/m K_mg/l LAL_µg/l Al MG_mg/l \\\n", "lab \n", "Eurofins 57.678245 29.250457 52.376600 29.250457 \n", "Eurofins (historic) 56.664388 34.381408 0.000000 25.563910 \n", "NIVA (historic) 97.547847 15.752093 0.052333 15.774522 \n", "VestfoldLAB (historic) 17.816004 17.816004 28.355407 17.816004 \n", "\n", " N-NO3_µg/l N N-TOT_µg/l N NA_mg/l P-TOT_µg/l P \\\n", "lab \n", "Eurofins 29.250457 29.250457 29.250457 29.250457 \n", "Eurofins (historic) 27.136022 25.085441 25.153794 25.017088 \n", "NIVA (historic) 15.737141 15.774522 15.774522 15.759569 \n", "VestfoldLAB (historic) 3.004009 17.460801 17.816004 16.207439 \n", "\n", " PH_ RAL_µg/l Al SIO2_µg/l Si SO4_mg/l \\\n", "lab \n", "Eurofins 99.908592 57.678245 29.250457 29.250457 \n", "Eurofins (historic) 93.438141 56.527683 0.000000 27.546138 \n", "NIVA (historic) 97.742225 36.326256 11.819677 15.759569 \n", "VestfoldLAB (historic) 99.873142 36.763587 0.000000 17.810930 \n", "\n", " TEMP_°C TOC_mg/l C \n", "lab \n", "Eurofins 20.566728 57.678245 \n", "Eurofins (historic) 27.614491 56.459330 \n", "NIVA (historic) 5.210825 15.774522 \n", "VestfoldLAB (historic) 18.176283 22.702593 " ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Percentage of total samples analysed per parameter, split by lab\n", "pct_df = df.groupby(\"lab\").agg(\"count\")\n", "tot_samps = pct_df[\"period\"].copy()\n", "\n", "for col in pct_df.columns:\n", " pct_df[col] = 100 * pct_df[col] / tot_samps\n", "\n", "pct_df = pct_df.iloc[:, 6:]\n", "\n", "pct_df" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0.5, 1.0, 'Percentage of total samples analysed per parameter, split by lab')" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot\n", "ax = pct_df.T.plot.bar(figsize=(15, 5))\n", "ax.set_ylabel(\"Percent of samples\")\n", "ax.set_title(\"Percentage of total samples analysed per parameter, split by lab\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## 4. Isolation Forests\n", "\n", "Isolation Forests are a type of random forest well suited to outlier detection. They have the advantage of making few assumptions about the underlying data distribution, which is useful in situations where - as here - most/all of the variables are strongly skewed.\n", "\n", "Isolation forests have a `contamination` parameter, which can be broadly interpreted as the \"expected proportion of outliers in the dataset\". In other words, setting `contamination=0.01` roughly translates to finding the most unusual 1% of data values. Without a strong theoretical basis for setting the `contamination` parameter, it must be found either by manual tuning or be fixed based on practical considerations (e.g. how many water samples can we realistically afford to reanalyse).\n", "\n", "### 4.1. `CA` and `PH` only\n", "\n", "The code below applies the isolation forest algorithm to `CA` and `PH`. Since these are almost always measured, this includes virtually all water samples (both new and historic) in the dataset." ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/opt/conda/lib/python3.10/site-packages/sklearn/base.py:450: UserWarning: X does not have valid feature names, but IsolationForest was fitted with feature names\n", " warnings.warn(\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "The total number of samples in the dataset is: 34852.\n", "\n", "The total number of outliers detected is 349:\n", " 320 in the 'historic' period\n", " 29 in the 'new' period\n", "\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
vannmiljo_codesample_datelabperioddepth1depth2CA_mg/lPH_<ubenevnt>pred
3930021-284502022-07-08Eurofinsnew0.00.07.76.7outlier
3931021-284502022-07-28Eurofinsnew0.00.07.66.8outlier
3933021-284502022-08-17Eurofinsnew0.00.012.06.8outlier
3935021-284502022-09-09Eurofinsnew0.00.08.66.6outlier
4871021-463732022-07-05Eurofinsnew0.00.06.17.2outlier
4872021-463732022-07-19Eurofinsnew0.00.07.97.4outlier
4873021-463732022-08-11Eurofinsnew0.00.07.47.4outlier
5752022-320182022-09-23Eurofinsnew0.00.09.67.5outlier
5847022-320192022-08-01Eurofinsnew0.00.05.47.0outlier
5849022-320192022-09-23Eurofinsnew0.00.07.46.9outlier
5940022-320202022-07-04Eurofinsnew0.00.07.46.8outlier
5941022-320202022-07-22Eurofinsnew0.00.06.07.0outlier
5942022-320202022-07-26Eurofinsnew0.00.06.26.8outlier
5943022-320202022-08-01Eurofinsnew0.00.06.56.7outlier
6243022-457692022-07-19Eurofinsnew0.00.06.56.6outlier
6245022-457692022-08-17Eurofinsnew0.00.05.96.5outlier
6246022-457692022-08-29Eurofinsnew0.00.05.76.5outlier
6762022-589042022-07-04Eurofinsnew0.00.06.06.7outlier
6763022-589042022-07-22Eurofinsnew0.00.05.76.8outlier
6764022-589042022-07-26Eurofinsnew0.00.05.46.8outlier
6765022-589042022-08-01Eurofinsnew0.00.05.66.8outlier
6770022-589042022-09-29Eurofinsnew0.00.05.86.8outlier
14226027-792782022-07-05Eurofinsnew0.00.04.87.1outlier
14227027-792782022-08-02Eurofinsnew0.00.04.97.2outlier
17196030-588382022-08-16Eurofinsnew0.00.07.36.2outlier
17197030-588382022-08-30Eurofinsnew0.00.05.66.4outlier
19593036-587512022-09-07Eurofinsnew0.00.011.07.9outlier
31621079-588782022-09-13Eurofinsnew0.00.06.66.2outlier
33482082-588702022-09-06Eurofinsnew0.00.010.07.3outlier
\n", "
" ], "text/plain": [ " vannmiljo_code sample_date lab period depth1 depth2 CA_mg/l \\\n", "3930 021-28450 2022-07-08 Eurofins new 0.0 0.0 7.7 \n", "3931 021-28450 2022-07-28 Eurofins new 0.0 0.0 7.6 \n", "3933 021-28450 2022-08-17 Eurofins new 0.0 0.0 12.0 \n", "3935 021-28450 2022-09-09 Eurofins new 0.0 0.0 8.6 \n", "4871 021-46373 2022-07-05 Eurofins new 0.0 0.0 6.1 \n", "4872 021-46373 2022-07-19 Eurofins new 0.0 0.0 7.9 \n", "4873 021-46373 2022-08-11 Eurofins new 0.0 0.0 7.4 \n", "5752 022-32018 2022-09-23 Eurofins new 0.0 0.0 9.6 \n", "5847 022-32019 2022-08-01 Eurofins new 0.0 0.0 5.4 \n", "5849 022-32019 2022-09-23 Eurofins new 0.0 0.0 7.4 \n", "5940 022-32020 2022-07-04 Eurofins new 0.0 0.0 7.4 \n", "5941 022-32020 2022-07-22 Eurofins new 0.0 0.0 6.0 \n", "5942 022-32020 2022-07-26 Eurofins new 0.0 0.0 6.2 \n", "5943 022-32020 2022-08-01 Eurofins new 0.0 0.0 6.5 \n", "6243 022-45769 2022-07-19 Eurofins new 0.0 0.0 6.5 \n", "6245 022-45769 2022-08-17 Eurofins new 0.0 0.0 5.9 \n", "6246 022-45769 2022-08-29 Eurofins new 0.0 0.0 5.7 \n", "6762 022-58904 2022-07-04 Eurofins new 0.0 0.0 6.0 \n", "6763 022-58904 2022-07-22 Eurofins new 0.0 0.0 5.7 \n", "6764 022-58904 2022-07-26 Eurofins new 0.0 0.0 5.4 \n", "6765 022-58904 2022-08-01 Eurofins new 0.0 0.0 5.6 \n", "6770 022-58904 2022-09-29 Eurofins new 0.0 0.0 5.8 \n", "14226 027-79278 2022-07-05 Eurofins new 0.0 0.0 4.8 \n", "14227 027-79278 2022-08-02 Eurofins new 0.0 0.0 4.9 \n", "17196 030-58838 2022-08-16 Eurofins new 0.0 0.0 7.3 \n", "17197 030-58838 2022-08-30 Eurofins new 0.0 0.0 5.6 \n", "19593 036-58751 2022-09-07 Eurofins new 0.0 0.0 11.0 \n", "31621 079-58878 2022-09-13 Eurofins new 0.0 0.0 6.6 \n", "33482 082-58870 2022-09-06 Eurofins new 0.0 0.0 10.0 \n", "\n", " PH_ pred \n", "3930 6.7 outlier \n", "3931 6.8 outlier \n", "3933 6.8 outlier \n", "3935 6.6 outlier \n", "4871 7.2 outlier \n", "4872 7.4 outlier \n", "4873 7.4 outlier \n", "5752 7.5 outlier \n", "5847 7.0 outlier \n", "5849 6.9 outlier \n", "5940 6.8 outlier \n", "5941 7.0 outlier \n", "5942 6.8 outlier \n", "5943 6.7 outlier \n", "6243 6.6 outlier \n", "6245 6.5 outlier \n", "6246 6.5 outlier \n", "6762 6.7 outlier \n", "6763 6.8 outlier \n", "6764 6.8 outlier \n", "6765 6.8 outlier \n", "6770 6.8 outlier \n", "14226 7.1 outlier \n", "14227 7.2 outlier \n", "17196 6.2 outlier \n", "17197 6.4 outlier \n", "19593 7.9 outlier \n", "31621 6.2 outlier \n", "33482 7.3 outlier " ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Columns of interest\n", "key_cols = [\"vannmiljo_code\", \"sample_date\", \"lab\", \"period\", \"depth1\", \"depth2\"]\n", "par_cols = [\"CA_mg/l\", \"PH_\"]\n", "\n", "# Run algorithm\n", "data = df[key_cols + par_cols].dropna()\n", "data = utils.isolation_forest(data, par_cols, contamination=0.01)\n", "\n", "# Summarise results\n", "all_out = data.query(\"pred == 'outlier'\")\n", "his_out = data.query(\"(pred == 'outlier') and (period == 'historic')\")\n", "new_out = data.query(\"(pred == 'outlier') and (period == 'new')\")\n", "\n", "csv_path = os.path.join(fold_path, \"isoforest_ca_ph.csv\")\n", "new_out.to_csv(csv_path, index=False)\n", "\n", "print(f\"The total number of samples in the dataset is: {len(data)}.\\n\")\n", "print(\n", " f\"The total number of outliers detected is {len(all_out)}:\\n\"\n", " f\" {len(his_out)} in the 'historic' period\\n\"\n", " f\" {len(new_out)} in the 'new' period\\n\"\n", ")\n", "\n", "new_out" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This initial approach identifies the **strangest 1% of the dataset overall**. \n", "\n", " * Most of the unusual values (300 out of 326) are actually in the historic dataset (i.e. they are already in Vannmiljø). \n", " \n", " * The 26 samples in the 'new' dataset that have been classified as outliers are predominantly those with unusually high concentrations of `CA` (greater than around 5 mg/l; see plots below)." ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot all samples\n", "fig, ax = plt.subplots(figsize=(6, 6))\n", "sn.scatterplot(\n", " data=data,\n", " x=\"PH_\",\n", " y=\"CA_mg/l\",\n", " hue=\"pred\",\n", " ax=ax,\n", " hue_order=[\"outlier\", \"inlier\"],\n", ")\n", "_ = ax.set_title(\"All data\")" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkwAAAJNCAYAAADDOCpmAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOzdd5hU1f3H8fdtU7cvZZe+gFTpKCjYwC7BbqzBGmvUny0m9miiGE0xJtbErrH3Suy9oIj03heWZftOufX3x8rCMDO7wC5b2O/reXxk77lz7znDwHw459xzFM/zPIQQQgghRFpqa1dACCGEEKKtk8AkhBBCCNEICUxCCCGEEI2QwCSEEEII0QgJTEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBBCCNEICUxCCCGEEI2QwCSE2G0pisKBBx7Y2tVoU26++WYUReGjjz5q7aoI0a5IYBJiN7FixQoUReHmm29OOH7mmWeiKAqKovDWW2+lfO3mL9GHH364BWraMEVROPPMM1u7GvU2v69tqU5CiJYngUmIDuSaa67BcZzWroYQQrQ7EpiE6CD69+/P3Llz+c9//tPaVRFCiHZHApMQHcQNN9xAKBTixhtvpLa2drtfV1ZWxu9+9zsGDx5MMBgkOzubyZMn89577yWc9+6776IoCtdff33C8ffee69+SHD16tUJZSeddBKKorBs2bKdbpdpmtx6663069cPv99PUVER119/PfF4POX569at4w9/+AMTJkygoKAAn89Ht27dOOWUU5g7d27CuTfffDNFRUUAPPbYY/XtUBSFRx99tP7+9957L0ceeSS9e/fG7/eTm5vL5MmTefPNN3e4PRUVFVx77bUMGDCAQCBAbm4uhx56KDNmzEg696OPPqofhp01axZHHXUUOTk5hEIh9t9/fz7//PNG71deXk4oFKJfv354npfynClTpqAoCjNnztzh9gixu5DAJEQH0a1bN6688krWr1/PnXfeuV2vWblyJWPGjOGOO+6gS5cuXHjhhfzyl79k/vz5HH744Tz44IP15+633374fD7ef//9hGt88MEH9b/euszzPD766CP69OlD3759d6pNnudx0kknceONN6IoCpdccglTpkzhP//5DyeeeGLK13zyySfccccd5OTkcPzxx3P55Zczbtw4XnjhBfbee29mzZpVf+6BBx7IZZddBsCIESO46aab6v8bOXIkUBcoL7vsMqqrqznkkEO44oorOProo/n++++ZMmVKwnvUmPLycvbZZx+mT59Obm4ul19+OccffzxfffUVhx12GP/6179Svu67775j3333JRaLce655zJlyhQ+//xzJk+ezPz58xu8Z25uLieffDLLli3jf//7X1L56tWrefvttxkzZgxjxozZ7rYIsdvxhBC7tWnTpnmAN2PGDK+6utrr2rWrFw6HvXXr1tWfc9NNN3mA99BDDyW89oADDvAURfGee+65hOPl5eXeiBEjvEAg4BUXF9cf32+//TxN07zKysr6Y2PHjvXGjh3r5ebmeqeffnr98VmzZnmAd/bZZ+9025566ikP8MaPH+9Fo9H645s2bfL69u3rAd4BBxyQ8JoNGzZ4VVVVSdeaOXOmFwqFvMMOOyzh+PLlyz3AmzZtWso6xGIxb/Xq1UnHy8rKvMGDB3u5ubleJBLZrvacd955HuBdeOGFCccXLFjgZWZmeoZheMuWLas//uGHH3qAB3iPPvpowmvuv/9+D/AuuOCChOObf68//PDD+mPfffedB3jHH398Up1uuOEGD/AefPDB7WqDELsr6WESogPJyMjglltuoba2lhtuuKHBc3/88Uc+/vhjTjjhhKTempycHG655RZisRgvvvhi/fHJkyfjOA4ff/wxUDe89P3333PooYdy4IEHJvQwbf715MmTd7o9jzzyCAB/+tOfCAQC9cfz8vLStq9Lly5kZmYmHR89ejSTJk3io48+wrKs7a6D3++nR48eScdzc3M555xzKC8v59tvv230OqZp8uSTT5KRkcEf//jHhLKBAwfym9/8BsuyeOKJJ5JeO3HiRKZNm5Zw7Oyzz0bX9e2695gxY9hrr7147bXX2LBhQ/1xx3H497//TWZmJqeeemqj1xFidyaBSYgO5txzz2XIkCE88sgj/PTTT2nP+/LLL4G60HPzzTcn/bd5DtOCBQvqXzNp0iRgyzDcRx99hOu6TJ48mcmTJ1NcXFw/RLT5nM2v2Rnff/89qqoyceLEpLKG1l968803+cUvfkFhYSGGYdTPS3rjjTeIx+OUlpbuUD3mzp3LmWeeSd++fQkGg/XXu+qqqwBYu3Zto9dYuHAh0WiUkSNHkpubm1R+8MEHA3Vt3tbYsWOTjhmGQdeuXSkvL9+uNlx00UVYlpXwUMDrr7/OunXrOP300wmHw9t1HSF2V3prV0AI0bI0TePOO+9kypQpXHPNNbz99tspz9u0aRMAM2bMSDnheLOampr6X48fP55wOFzfe/T+++8TCATYd9996datW/2xPfbYg08++YQhQ4ZQUFCw022prKwkLy8PwzCSytJd95577uGyyy4jNzeXQw45hF69ehEKhVAUhVdeeYUff/wx7YTxVL766ismTZqEbdtMnjyZqVOnkpWVhaqqzJo1i1dffXW7rldZWdlgvQsLCxPO21p2dnbK1+i6vt3LSPzyl7/kyiuv5OGHH+baa69FURQeeOABAC644ILtuoYQuzMJTEJ0QEcddRSTJk3inXfeSRuGNn8J//3vf+fSSy/drusahsHEiRN57733KCkp4f3332fChAkEAgEGDRpEt27d+N///sfo0aOprq5uUu/S5jqWlZVhWVZSaFq/fn3S+bZtc9NNN1FQUMD3339fH0I229yrtiNuu+02otEoH374YVKv1u23386rr766XdfZ/H6nqjdAcXFxwnnNLRgMcuaZZ/KXv/yFGTNmMGDAAN577z3Gjx/P8OHDd8k9hWhPZEhOiA7qrrvuQlEUrr76alzXTSofP348AJ9++ukOXXfy5Ml4nsfTTz/N/PnzE+YoTZ48mY8++qg+pDVl/hLUzTtyXZfPPvssqSzV1h+lpaVUVFSw7777JoWlmpqalMNdmqYBpO2pWbJkCXl5eSmHADfP5doeAwcOJBQKMWvWrJTDaB9++CFQ1+Zd5cILL6zvWXrooYdwXZfzzz9/l91PiPZEApMQHdSoUaM4/fTT+fHHH3nmmWeSyseOHct+++3HSy+9lHaxy59++omSkpKEY5t7jW6//XaApMBUWVnJv/71L1RVbfI+b2eddRYA1113HbFYrP54WVkZt912W9L5Xbp0IRQK8d133yUMJVqWxWWXXZZy7lJubm7KNaQ269OnD2VlZcyePTvh+L///W/efffd7W6Lz+fjtNNOo6amhhtvvDGhbOnSpdxzzz0YhsEZZ5yx3dfcUf379+eQQw7htdde48EHHyQnJ4df/vKXu+x+QrQnMiQnRAf2xz/+keeff54lS5akLH/66aeZNGkS55xzDvfccw/jxo0jJyeHNWvWMHv2bObMmcOXX35Jly5d6l8zatQo8vLyKCkpITs7O2Htns3hqaSkhLFjx5KTk9Ok+p9yyik8++yzvPbaa+y5554cffTRWJbFCy+8wF577cXSpUsTzldVlUsvvZQ77riDYcOGcfTRR2OaJh9++CFlZWUcdNBB9T05m2VkZDBu3Dg++eQTTj/9dPbYYw80TWPq1KkMHz6cyy+/nHfffZeJEydy0kknkZ2dzXfffcdnn33GCSecwAsvvLDd7bnjjjv49NNPuffee/n222856KCDKC0t5bnnnqO6upp77723fiHNXeXCCy/kvffeo7S0lEsvvZRgMLhL7ydEeyE9TEJ0YD179uTyyy9PW96jRw9mzpzJH//4RzRN46mnnuKee+7hiy++oFevXjzwwAMMGzYs4TVb9xwdcMAB9UNam683YMAAoGlPx22mKArPP/88t9xyC67rcu+99/Laa69x1lln8dxzz6V8za233srdd99NMBjkgQce4KWXXmLs2LF888039OrVK+VrnnjiCY466ijefvttbrnlFm644Yb64bvDDz+c119/nSFDhvDss8/y73//G7/fz4cffshRRx21Q+3Jy8vjyy+/5JprrmHTpk385S9/4fnnn2fvvffmnXfe4aKLLtqxN2gn/OIXv6BTp04AMhwnxFYUz0uzFr4QQogOZ+nSpeyxxx5MnDiRTz75pLWrI0SbIT1MQggh6v35z3/G8zwuueSS1q6KEG2K9DAJIUQHt3LlSp544gkWL17ME088wciRI/nuu+9QVfk3tRCbSWASQogO7qOPPuKggw4iHA6z3377cd9999GnT5/WrpYQbYoEJiGEEEKIRkh/qxBCCCFEIyQwCSGEEEI0QgKTEEIIIUQjOtxK3+Xl5di23eh5nTt3ZuPGjS1Qo7ano7Zd2t2xSLs7lo7abui4bd/eduu6Tm5ubuPnNUel2hPbtrEsq8FzFEWpP7ejzYnvqG2Xdku7OwJpd8dqN3Tctu+KdsuQnBBCCCFEIyQwCSGEEEI0QgKTEEIIIUQjJDAJIYQQQjSiw036boht20QiEQCi0SimabZyjVpHS7Y9FAqh6/IxFEII0bbJN9XPbNumtraWzMxMVFXFMIxGn6bbXbVU213Xpbq6mnA4LKFJCCFEmyZDcj+LRCL1YUm0DFVVyczMrO/VE0IIIdoqSQdbkbDU8uQ9F0II0R7It5UQQgghRCMkMAkhhBBCNEICkxBCCCFEIyQwdVDjxo3joYceau1qCCGEEO2CBCYhhBBCiEZIYGrHOurCmkIIIURLk8DUhpxwwglcd911XHfddQwePJihQ4cyffp0PM8D6obR/va3v3H55ZczaNAgrr76agC+/fZbjjvuOPr168fYsWO54YYbEtY2Ki0tZdq0afTr14/x48fz0ksvtUr7hBBCiPZKAlMb8/zzz6NpGq+//jq33norDz30EE8//XR9+f3338+gQYN4++23ufzyy5k/fz6nnXYaRxxxBDNmzOC+++7jm2++4brrrqt/zf/93/+xZs0ann32WR588EEee+wxSktLW6N5QgiRwOfzEfQZ+P3+1q6KEA2S/SjamG7dunHLLbegKAr9+/dnwYIFPPTQQ5x22mkATJgwgQsuuKD+/EsvvZRjjjmG8847D4C+ffty6623cvzxx3P77bezdu1aPvjgA15//XVGjx4NwN13380BBxzQ8o0TQoif6ZpGlhkl9v47WAt+Qu/Vl7xJR1LjD2O6bmtXT4gkbSIwzZs3j9dee43ly5dTXl7OVVddxd577w3U7fH23//+lx9++IGSkhJCoRDDhg3j1FNPJS8vr5Vr3vxGjx6Noij1P48ZM4YHHngAx3EAGD58eML5P/30EytWrODll1+uP+Z5Hq7rsnr1apYtW4au64wYMaK+vH///mRnZ+/ilgghRGqappFZVkLpdRfixWN1B7/5lJqXnyTvhrvxevbHktAk2pg2EZji8Th9+vThoIMO4u67704oM02T5cuXc/zxx9OnTx9qamp47LHHuPPOO7njjjtaqcatJxQKJfzsui6nn346Z599dtK53bt3Z+nSpQAJIUwIIVpT2DGpuPuGLWFpM8eh/M/Xk/fXx6lQ2sTXkxD12sQnctSoUYwaNSplWSgU4oYbbkg4dtZZZ/H73/+e0tJSOnXqlPJ1lmVhWVb9z4qiEAwG63/dVn3//fdJPxcVFaFpWsrzhw0bxsKFCykqKkpZ3r9/f2zb5scff6x/j5csWUJlZWXzVryJWvv3ZPP9W7seLU3aLe1uDVptLXbxmpRlXm0NlG9CyS9otvu1lXa3ho7a9l3R7jYRmHZUJBJBUZSk3patvfzyy7zwwgv1PxcVFTF9+nQ6d+6c8vxoNIphGAnHtv15V1MUhXXr1vGHP/yBadOmMXv2bP7zn/9wyy23YBgGiqKgaVpCvS699FKOPPJIrr/+ek4//XTC4TCLFi3i448/5vbbb2fw4MFMmjSJa665hrvvvhtd17n++usJBoNJ19paS7bd5/NRWFjYYvdrSEFB8/0l3Z5IuzuW1m63ubSqwXLVdXbJ3wmt3e7W1FHb3pztbneByTRNnn76aSZMmNBgYDr22GOZMmVK/c+bU+bGjRuxbTvldbfukTIMI+HnluB5HieccAKRSITDDjsMTdM4++yzOeWUU7AsC8/zcBwnoV4DBgzghRdeYPr06UydOhXP8+jduzdTp06tP+/uu+/mqquu4phjjqFTp05cc801rF27Nulam7V0203TpLi4uMXul4qiKBQUFLB+/fr6ZRw6Amm3tLs1ZAeCqJnZuNUperp1HfI6N+vfCW2l3a2ho7Z9R9qt63razpSE85qrci3Btm3+9re/4Xke5557boPnGoaRtpekLX9odF3nD3/4Q8r5WV9//XXK14wcOZJnnnkm7TW7dOnC448/nnDshBNOaFpFm1lb+T3xPK/N1KUlSbs7ltZud60RIOv8q6i464akssxTf01E9+G5zV+/1m53a+qobW/OdrebdZhs2+avf/0rGzdu5Prrr2+wd0kIIUTbZdk29qAR5N/+AP49R6NmZmP0H0zuDXej7n8YsV0QloRoqnbRw7Q5LK1fv56bbrqJzMzM1q6SEEKIJoh6EO/SneD/3UzYsXFUjVpVr19CRYi2pk0Eplgsxvr16+t/LikpYcWKFWRkZJCbm8tf/vIXli9fzm9/+1tc16WiogKAjIwMdL1NNKFZbD1JXQghdneu61ILoOjgARKWRBvWJtLG0qVLueWWW+p/3jzf5oADDuDEE0/ku+++A+Caa65JeN1NN93E0KFDW66iQgghhOiQ2kRgGjp0KM8991za8obKhBBCCCF2tXYz6VsIIYQQorVIYBJCCCGEaIQEJiGEEEKIRkhgEkIIIYRohAQmkVL37t155513AFi9ejXdu3dnzpw5rVwrIYQQonVIYNpF2ssS9HfffTeHHHJIg+d069aNH374gUGDBrVQrYQQQoi2pU0sK7C78GIR3JefhB+/AccGTYcRe6MeezpKoP1u5aJpGl26dGnSNUzTxOfzNVONhBBCiJYlPUzNxItFcG+/Bj58EzaVQEVZ3f8/egv39mvwYpFdct94PM4NN9zA8OHD6du3L8cccwyzZs0C4Nlnn2Xw4MEJ57/zzjt07969vvwvf/kL8+bNo3v37nTv3p1nn3026R6phuQWLVrEGWecwR577MGIESP4zW9+Q1lZWX35CSecwHXXXcfNN9/MnnvuySmnnLILWi+EEEK0DAlMzcR9+UkoXg3bDsW5LqxfU1e+C/zxj3/krbfe4m9/+xvvvPMOffr04bTTTqO8vLzR106dOpXzzz+fgQMH8sMPP/DDDz8wderURl+3YcMGjj/+eIYMGcLbb7/NU089RWlpKeeff37Cec8//zy6rvPKK68wffr0nW6jEEII0dpkSK65/PhNcljazHXryk/5dbPeMhKJ8Pjjj/PXv/6VSZMmAfDnP/+Z8ePH89///pe8vLwGXx8MBgmHwzs85Pb4448zbNgwfve739Ufu/vuu9lrr71YunQp/fr1A6BPnz5cf/31O9EyIYQQom2RwNQMPM+rm7PUEMfB8zwURWm2+65YsQLLsthrr73qjxmGwciRI1m8eDHjxo1rtnttbfbs2XzxxRfsscceSWUrV66sD0wjRozYJfcXQgghWpoEpmagKErdBO+GaFqzhiXY8iTettfdHMxUVU16Ws+yrGa57yGHHMLvf//7pLKuXbvW/zoYDDb5XkIIIURbIHOYmsuIvUFN83aqal15MysqKsLn8/HNN9/UH7Msix9//JE99tiD/Px8ampqiES2TDifO3duwjUMw8B13R2675577snChQvp2bMnRUVFCf+FQu33aUAhhBAiHQlMzUQ99nQo6JEcmlQVCnrUlTezUCjEGWecwW233caHH37IokWLuPrqq4nFYpx88smMGjWKYDDIHXfcwfLly3n55Zd5/vnnE67Rs2dPVq1axZw5cygrKyMejzd63zPPPJOKigouuugifvjhB1auXMnHH3/MFVdcgeM4zd5OIYQQorVJYGomSiCE+rs74cAjIb8L5OTX/f/AI1F/d+cuW4fp97//PUceeSSXXnophx9+OCtWrOCpp54iJyeH3Nxc/vGPf/D+++9z8MEH88orr3DFFVckvP7II4/kwAMP5KSTTmLYsGG88sorjd6zoKCAV155Bdd1Oe2005g0aRI33ngjmZmZqOl62YQQQoh2TPHay5LUzWTjxo0p5/FUVVWRlZVV/7NhGE2a79PcE7xbUlPbvqO2fe9bg6IoFBYWUlxc3G5WaW8O0m5pd0fQUdsNHbftO9JuwzDo3Llzo9eU7oBdpL2GJSGEEEIkk8AkhBBCCNEICUxCCCGEEI2QwCSEEEII0QgJTEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBBCCNEICUxCCCGEEI2QwNTOnXDCCdx4443bde7q1avp3r07c+bMAeCLL76ge/fuVFZW7soqCiGEEDtNVVV0XW/1BaH1Vr37bqyltkZ56KGHMAxjp147duxYfvjhh1bflkQIIYTYlq5pZFpxnOWLcUpLMPoPxs3rRDVaq2zzIoGpGUUsh6d+3Mg3a2qxXQ9dVdi7R5jTRnQmZGi75J65ubk7/Vqfz0eXLl2adH/TNPH5fE26hhBCCLE1XdPIKFnLppsvw4tF648b/QaRc+0dlLfCAJkMyTWTiOXw23dX8ubCCkpqLcqiNiW1Fm8trOC3764kYjm75L5bD8mNGzeOe+65hyuuuIIBAwaw11578eSTT6Z9baohuW+//ZapU6fSr18/xo4dyw033EAkEqkvHzduHH/729+4/PLLGTRoEFdfffUuaZcQQoiOK9OOU3bL5QlhCcBauoDapx8gqEtgaree+nEjqytNtu0kdIE1lSZP/bixRerxwAMPMHz4cN59912mTZvG7373O5YsWbJdr50/fz6nnXYaRx11FDNmzOC+++7jm2++4brrrks47/7772fQoEG8/fbbXH755bugFUIIIToqVVVxVi3Di0ZSlkc/eQ9/PNbCtZLA1Gy+WVObFJY2c38ubwmTJk3izDPPpKioiIsvvpi8vDy++OKL7XrtfffdxzHHHMP5559P37592Wuvvbj11lt54YUXiMW2fDgnTJjABRdcQFFREUVFRbuqKUIIITogVVVxNjXQyeA4YFstV6GfyRymZuB5Hrbb8AQ0x/VaZCL4kCFD6n+tKAqdO3dm06ZN2/Xan376iRUrVvDyyy/XH/M8D9d1Wb16NXvssQcAw4cPb95KCyGEED9zHAej36C05WpuPp7P34I1qiOBqRkoioKuNhyENFVpkafmdD3xt1RRFFzX3a7Xuq7L6aefzq9//Wts204o6969e/2vQ6FQ0ysqhBBCpOB5Hm5OLr7BwzHnz04qz5p2MRFfEKyW7WWSwNRM9u4R5q2FFaSKJurP5W3dsGHDWLhwIX379sVq4Q+iEEIIsVk1GjlX3krkuUeIfPAm2BZafmcyf3Ux7tBRrfIdJYGpmZw2ojOz10dYU2kmhCYV6JHt47QRnVuratvtoosu4he/+AW//e1vOfnkkwmFQixZsoRPPvmE2267rbWrJ4QQooPwPI9yFAKnnEf+cWeAbeH5fNQagaQRkJYigamZhAyN6Yf1rl+HyXE9tBZYh6k5DRkyhBdffJE777yT4447Ds/z6N27N1OnTm3tqgkhhOiAYrZDTDNA+3mB5lYKSwCK1xrLZbaijRs3puzKq6qqSljx2jCMJnX5tdRK37tCU9u+o7Z971uDoigUFhZSXFzcKivIthZpt7S7I+io7YaO2/YdabdhGHTu3PgokCwrsIu017AkhBBCiGQSmIQQQgghGiGBSQghhBCiERKYhBBCCCEaIYHpZx1pMlxbI++9EEKItk4C0890Xae2tla+vFuQ53nU1tYmrU4uhBBCtDXyTfWzcDhMPB6nuroaAJ/Ph2marVyr1tGSbff7/fj9Lb8nkBBCCLEjJDBtZfOXd0ddtwI67podQgghRENkSE4IIYQQohESmIQQQgghGiGBSQghhBCiERKYhBBCCCEaIYFJCCGEEKIREpiEEEIIIRohgUkIIYQQohESmIQQQgghGiGBSQghhBCiERKYhBBCCCEaIYFJCCGEEKIREpiEEEIIIRohgUkIIYQQohESmIQQQgghGiGBSQghhBCiERKYhBBCCCEaIYFJCCGEEKIREpiEEEIIIRohgUkIIYQQohF6a1dACCGE6EiCmkrAjOFVV6L4/DihDGo1A8dxWrtqogFtIjDNmzeP1157jeXLl1NeXs5VV13F3nvvXV/ueR7PP/8877//PjU1Neyxxx6cc8459OzZsxVrLYQQQuyYLNXDfP0ZSl/7L/wckPQevcm99g4qQ1kSmtqwNjEkF4/H6dOnD2effXbK8ldffZU333yTs88+m9tvv52cnBxuu+02otFoC9dUCCGE2Dk+Q8f5/ANqX36qPiwB2GtWsun6i8m0zVasnWhMmwhMo0aN4uSTT2bcuHFJZZ7n8dZbb3Hssccybtw4evXqxcUXX0w8Huezzz5rhdoKIYQQOy5oxqh54dGUZW5FGe6qZahqm/haFim0iSG5hpSUlFBRUcGIESPqjxmGwZAhQ1i4cCGHHHJIytdZloVlWfU/K4pCMBis/3VDNpc3dt7uqKO2Xdot7e4IpN2t227VsnGrKtOW26uWou8xNOG7q6naSttb2q5od5sPTBUVFQBkZ2cnHM/Ozqa0tDTt615++WVeeOGF+p+LioqYPn06nTt33u57FxQU7FhldyMdte3S7o5F2t2xtHa77dINqFnZaUOTv2gPcjp12iX3bu22t5bmbHebD0ybbZsSPc9r8Pxjjz2WKVOmJL1+48aN2Lbd6L0KCgpYv359o/fZ3XTUtku7pd0dgbS7ddttGDoZx0+j6pF7ksrU7Fzo3ofi4uJmvWdbaXtL25F267q+XZ0pbT4w5eTkAHU9Tbm5ufXHq6qqknqdtmYYBoZhpCzb3g+N53kd6gO2tY7adml3xyLt7lhau92maRGYcDDh8lJqX392y1Ny3X9+Sk734e2ip+Rau+2tpTnb3eYDU5cuXcjJyWH27NkUFRUBYNs28+bN47TTTmvl2gkhhBDbr8qD4NRT6HT4cVutwxSmUvPJkgJtXJsITLFYjPXr19f/XFJSwooVK8jIyKBTp04ceeSRvPzyyxQWFlJQUMDLL7+M3+9n4sSJrVhrIYQQYsdFHY+o5oOcrYaBJCy1eW0iMC1dupRbbrml/ufHH38cgAMOOICLL76Yo48+GtM0efjhh6mtraV///5cd9119U+9CSGEEELsSm0iMA0dOpTnnnsubbmiKJx00kmcdNJJLVgrIYQQQog6skKWEEIIIUQjJDAJIYQQQjRCApMQQgghRCMkMAkhhBC7GV3XycIlx7GwN5Vg6G1iynK7Ju+gEEIIsRsJKx7aop+ofuyf2OtWoebkkXHCrwiOP4hKr2PtKdecpIdJCCGE2E34dB1l1teU3/5b7HWrAHAryqh6+G/EnvsPYVUC086SwCSEEELsJkJWjKpH/5GyLPLeq/jikRau0e5DApMQQgixu6itwaupTlvsrFudtJm92D4SmIQQQojdhJJm0/n68nBGC9Vk9yOBSQghhNhNWIEQvoFDU5apGVko+V3wPK+Fa7V7kMAkhBBC7CYiikb2ZTei5uYnHFd8fnKvv4sa3ddKNWv/ZFkBIYQQYjfhui6VgQxypz+Es2wh1vyf8PXuiz50FNVGANtxWruK7ZYEJiGEEGI34jgOFYqONmA4+pDRZOTnU1xcjCdhqUkkMAkhhBC7IcdxcF23taux25A5TEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBBCCNEICUxCCCGEEI2QwCSEEEII0QgJTEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBBCCNEICUxCCCGEEI2QwCSEEEII0QgJTEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBBCCNEICUxCCCGEEI2QwCSEEEII0QgJTEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBBCCNEICUxCCCGEEI2QwCSEEEII0QgJTEIIIYQQjZDAJIQQQgjRCAlMQgghhBCNkMAkhBAipYCuk+G5hFUFVZWvC9Gx6a1dASGEEG2LoapkxiNEXn6O2p9mouXmkXncr3B7FFHttXbthGgdEpiEEELUU1WVjKoySn97Hp4ZB8BevZz47JmEjzud0JEnEZHQJDog6WMVQghRL+Q5VP3rjvqwtLXal57EH4+2Qq2EaH0SmIQQQtTTzTjm4nlpy615s9B1GZwQHY8EJiGEEPWURk+Qrw3RMcknXwghRD3LF8A3eETacmPICGzbbsEaCdE2SGASQghRL6KoZF1wNUowlFSWeer5xHyBVqiVEK1PBqKFEELUc12X6swcOv31caLvv4E5+1vU3E6EjzkVq1OBPCEnOiwJTEIIIRLYjkuZ5sN35EkEDjsOV9OodFw8T9KS6LgkMAkhhEjJtG1MANtp7aoI0epkDpMQQgghRCMkMAkhhBBCNEICkxBCCCFEIyQwCSGEEEI0QgKTEEIIIUQjJDAJIYQQQjRCApMQQgghRCMkMAkhhBBCNEICkxBCCCFEIyQwCSGEEEI0QgKTEEIIIUQjJDAJIYQQQjRCApMQQgghRCP01q7A9nIch+eff55PP/2UiooKcnNzOfDAAznuuONQVcl9QgghhNh12k1gevXVV5kxYwYXX3wxPXr0YNmyZfzrX/8iFApx5JFHtnb1hBBCCLEbazeBadGiRYwdO5bRo0cD0KVLFz777DOWLl3ayjUTQgghxO6u3QSmQYMGMWPGDNatW0e3bt1YsWIFCxcuZNq0aSnPtywLy7Lqf1YUhWAwWP/rhmwub+y83VFHbbu0W9rdEUi722e7FUUhoIDfjIHr4vj8RDQDx3G267Vb/7+j2BXtVjzP85rtaruQ53k888wzvPrqq6iqiuu6nHzyyRx77LEpz3/uued44YUX6n8uKipi+vTpLVVdIYQQolnY69dS/p97iH35EbgORp/+5F70W4z+g1H9gdauXofRbgLT559/zpNPPsnpp59Oz549WbFiBY8++ii/+tWvOPDAA5POT9fDtHHjRmzbbvBeiqJQUFDA+vXraSdvT7PpqG2Xdku7OwJpd/trd45nU/77C3A2bkgsUFU63fUfqnI647pu2te357Y3xY60W9d1Onfu3Og1282Q3JNPPsnRRx/NhAkTAOjVqxcbN27klVdeSRmYDMPAMIyU19reD43neR3qA7a1jtp2aXfHIu3uWNpbuzVNw543JzksAbgu1Y/9k8ClN1C7HU1qb21vLs3Z7nbzPH48Hk9aPkBV1Q75ARBCCLH7MwyD+LefpS2Pz/sRYzvmMYnm0W56mMaMGcNLL71Ep06d6NGjBytWrOCNN97goIMOau2qCSGEEM3OdV20zgVpy7WcXNyONZe7VbWbwHT22Wfz7LPP8vDDD1NZWUleXh6HHHIIJ5xwQmtXTQghhGh2pmmSe8Dh1LzwWMry8DGnETUCsNV8XbHrtJvAFAwGOfPMMznzzDNbuypCCCHSUFUVRVG265F30bhoMEzOZTdS8Y/bYKvJ3YHx+2OMP5BaCUstpt0EJiGEEG2XX1UJ2XGcZUvw4jH0foMw/cHtmpAs0ot54A0bS+d/PYc5dxZubTX+YWOwM7OpSP9wnNgFJDAJIYRokoCqYMyfRenfboatlm0JHXE8WSdMo0q+2Jsk7kFc86GP3heAqOPIA0+toN08JSeEEKJtCkZqqLj7hoSwBBB5+0W8uT+g6/Jv8+Zg2za2bUtYaiUSmIQQQuw0v99P9P3XIc2XeM0LjxG0zbSv13WdQCCQdt08IdoKif1CCCF2muJ5uOvXpi13yjaiei4oif8+1zSNLCuG9e0XWAt+wtenH5l770+1L4DlyBieaHskMAkhhNhpjufhG7E30a8+Tlnu22MItmaAu6UHStM0sqrL2PT7C/Fqa7ac/MT95N9yDzUFPbAlNIk2RobkhBBC7DTLsvCN2Qc1Kzu5UFHIPO0CoiSurhh2LCruuiExLAFYJuV3/JaMBobwhGgtEpiEEEI0SbURIP9P9+MbMqL+mFbQnbyb/04kp1PS5rBatBZ79YqU13KrKqGyfFdWV4idIkNyQgghmsS2bSqCmYSu+ANZZhzPcXADAWp1P/Y2T84BeCmOJZTHYyiKIk+DiTZFApMQQogmc12XGhQwArD5gbd0wSicgRLOSB6SA1A11E5dJSyJNkeG5IQQQrSoiC9A1lmXpizLOO50Yoa/hWskROOkh0kIIUSLMm0HY+R48v90H+76tagZmXjRCGTl4PbqR7WbundJURQCqoLfrts/La77iLluh+iN8us6QSsOCliqTsSjQ7S7LZHAJIQQolW4tTVUv/wU9urlaF0KyTz5XJSefWGbp+qgbimCbDNKzX8fZtNn7wMQ3P8Qck86m0pfcLfd7FdRFHJwiL76JGXvvooXj+IfNY6caZdQE87GcmX5hZYiQ3JCCCFalE/XUGZ+Tvkfr8ZevRwAp6SYintuxXzzeYJqcmDKtmNsuuZcoh+9A7YFtkX0g7fYdO2vybbjLd2EFpPtOVTceiW1rzyNF60F1yU+80tKrzyLzEgVipL8XoldQwKTEEKIFhUy41Q9dm/KstpXnyFgJQYgn2EQ/eBt3KqKpPPdijJin76Hz9j9Bkw0TcNdsRh75dLkQsuk+qkHCMq3eIuRt1oIIZqRrutk4ZIbqyE3HiFT8dA0rbWr1bbUVuNFalOXuQ5u6YaEnhOfFSf+deqVxAFiX36M8fO8pt2JYRjEv/ggbXn8+6/wNbJEg2g+u18kF0KIVuJTVULFq6i894/Y61YDYPTpT/Zvrqc6tzP2bjrPZkcpjWy0q/gDCROaPVVDCYXTnq+GwniqBrvZdiqe56Fk5aQtV8IZyLTvliM9TEII0QwURSEjUsWmGy6uD0sA1oolbPr9+WTFI61Yu7bFCYQwevdLWaZm5UB2bsKxmKISPvqUtNcLHX0K8RQTxdu7eDxO8KAj05aHjziBqCzB0GIkMAkhRDMIqCo1Lz4OKXqRvHic6Ptv4NOlUx+gVjPIvvIPqBlZCccVv5/c399JjZ4YAhzHgb6DCO53SNK1ggceAb3777ZPycUzssg848Kk476BQ/FPPhJThuRajPzpFUKIZqDbJrFFc9OWm3Nn4T/ieGRb2boAVJWRQ97dj2DN+xF74U9ovfvjGzWOGl8w5dBllaeQOe0SQlNPJvbpDFAUAvsdgpOTT5W3+/UubRbxFEIHHEGnvSYS++x/eDXV+Pc5EAp6UOGpIINyLUYCkxBCNANX09E6dUkYjtua1rUQV9PBlR4BqAtN5aqBPnI82pgJmK5LrWWl7KHbrNpTUHK74DvhLAAqTbNDLN4Y8SASzMR31C9RFIVqy6rb0LgDtL0tkSE5IYRoBlEUwiecmbY8/IuTidu757BRU9i2TTwex7K27yk3z/OIx+PE4/EOEZa2Zpom8Xi8LiyJFieBSQghmoHrujjdepF52vmgbvVXq6aRfdG1xHLyOtwXvBC7ExmSE0KIZlLjKQQPmkKniQdjr1yKomlovfoS0X3E0+yPJoRoHyQwCSFEM4p6HlHdj7rHnkBdzxMSloRo9yQwCSHELiDzTITYvcgcJiGEEEKIRkhgEkIIIYRohAQmIYQQQohGSGASQgghhGiEBCYhhBBCiEZIYBJCCCGEaIQEJiGEEEKIRkhgEkIIIYRohAQmIYQQQohGSGASQgghhGiEBCYhhBBCiEZIYBJCCCGEaMR2b7578cUXoyjKdp2rKAr/+Mc/drpSQgghhBBtyXYHpiFDhmx3YBJCCCGE2J3sUA+TEEIIIURHtN1zmB599FEWLFiwK+sihBC7DVVVUVWZJirE7mK7e5jmz5/P22+/TXZ2NnvvvTfjx49nyJAh8heCEEJsJahCwIxjL1+MomlovfsR0f3EPa+1qyaEaILtDkzTp0+npKSEr776iq+//poZM2aQkZHB2LFjGT9+PMOGDUPXt/tyQgix28lQPNz336T0vw+D69Yd1DSyz78abfS+RCQzCdFu7VDC6dKlC1OnTmXq1KmUlZXVh6fp06cTCAQYPXo048ePZ+TIkfh8vl1VZyGEaHNUVUVbu5zKpx9MLHAcKv91B53ufpRodj6e9DQJ0S7tdJdQXl4eRx55JEceeSQVFRV88803fP311/z1r3/FMAxGjhzJFVdc0Zx1FUKINiuIR+0Lj6Ytr339v/jPvJSYZbdcpZpI0zR0XcfzPEzT3K7X+Hw+FEXBsizczb1sQuwGmmUMLScnh0MPPZRDDz2Umpqa+vAkhBAdherYOKUlacudkmJ0p32EJVVVyXItnEVzMX/4Cr1LIeF9DiLiDxF3U/eQhRUPX3UlsU9n4EUjZE6YBF27U6Vo0qsmdgvNPukoIyODSZMmMWnSpOa+tBBCtFm2bmAMHIa9bnXKct+QkdiqDm7bDk2KopBjxym7/mKcjevrj1c/9SC5V98GA4Ylhaaw4uG++zKlzz9af6z27RfxDR1Jzv/dQjmyhp9o/3YqMP3rX/9KW6aqKqFQiP79+7P33nvLRHAhRIcQcz1yjzud6CfvguMklCn+AMGDp1Bmt+2wBBBUFaofvichLAHgupTfdQOd/vkscW3LHFVFUTAqNrFpq7C0mTl3FuYXH2DsdxhWO2i7EA3ZqTQzd+5cIpEIkUgEVVXJzMykuroa13UJhUIAvPnmm3Tr1o2bbrqJnJyc5qyzEEK0OZ7nURPKIv+2f1F575+w164EwOg7gOxLrqPKF0oKUm2R34xR/c2nqQsdB3vxPLShY3B+bovP5yP63itpr1f7xnNkjT8QS9F2QW2FaDk7FZiuvPJK7rrrLs477zzGjx+Pqqq4rsuXX37JU089xRVXXIHjONx1110888wzXHjhhc1dbyGEaHNM18Xp2oPMm/+OGougKAp2IESVqtcHjLbOc+wtSyKk4NZUJ2yTpXgebnVl+utFalGQOUyi/dupVScff/xxfvGLX7DvvvvWL1ypqioTJkxgypQpPPbYYwwcOJCjjz6aWbNmNWd9hRCiTXMchypUKgIZlPvDVHtKuwlLAK4vgN6tV9py3+Dh2FsNr1mOg3/iwWnP94/dB1OXZWZE+7dTgWnp0qX06NEjZVnPnj1ZsWIFAH369KG6unqnKyeEEKJlRXQfWRdcDSk2Ww+M2w87IyvhmOM4aAOGonfvnXS+4g+QceLZxNI8WSdEe7JTgSkYDDJ37tyUZXPmzCEYDAJgmmb9r4UQQrR9tm0T79aH/DsewDdoGCgKam4+mWf+hvCvr6baSw5SVapBzs1/Izz1FJRgCFSNwLj96XT3I1QFM2RZAbFb2Kk5TBMnTuTVV1/F8zz22WcfsrOzqays5IsvvuD111/nyCOPBGDZsmV07969WSsshBBi14p5HmanboSu+iOZnoPnQcTwE0nzpJvrupSj4T/2dPKmnASAqeuUu+C1o+FIIRqyU4Hp1FNPpby8nFdeeYVXXnkloWzChAmccsopAAwYMICRI0c2tY5CCCFamOu61ACggQJsx7IAcdshvvlpOEd6lcTuZacCk67rXHbZZRx//PHMmzePmpoaMjIyGDJkSMLcpuHDhzdbRYUQQgghWkuTVpXs0aNH2snfQgghREP8uk7QjEI8Bj4/MV+QmAzhiTaqSYEpFotRWlqaclPGvn37NuXSQgghdmPZKpjvvEjZq0/jxaKgG4QPmUru8b+iPMXEciFa204FpqqqKu6//35mzpyZ9pxnn312pyslhBBi9xXQNOKvPUXtK09vOWhb1L79Ik75JsLnXkGtTIESbcxOBaYHH3yQuXPncuSRR9K9e3fZL04IIcR2C5pRSt94PmVZ7KuPyDjjQmp9siSNaFt2KunMmTOHM844g4MPTr+6qxBCCJGKF6kB20pb7pZtRCns3S7Wb9I0jQzHRK2pxo1GUHPzifsCRGSxzt3OTgUmv99P586dm7suQgghOgDFH2iwXM3IahdhSdd1sqrLKfvTNTjr19YdVBRChx5N9klnUenKXKzdyU6t9L3//vvz5ZdfNnddhBBCdACWP4R/2JiUZXphD9xttl9pqzKsGJuuv3hLWALwPCLvvoL1/hv4ZLrKbmWnAtPJJ5+Mbdv8+c9/5v333+frr79O+k8IIUTHo+s6hmE0eE4tClm/uQ69V+LT1FqnruRefxc1WsOv31FbbxLfnNd0Vy/HrapIWV7z6jOEzGiz3U+0vp2KvyUlJSxZsoTi4mK+++67lOfIU3JCCNFxBBQIWjHMH77Hi0XJGDYGKzObmhRLBHieR4XmI+uGv6BUbMJetwqtSyFKflcqdR9OM63FpGkaWXYcZ8kCqj59l4w+/VG69aRKNXBdt8nXttesTFvu1dbUrY7ua97wJ1rPTj8lF4lEmDZtGj169Gixp+TKysp48sknmTVrFqZpUlhYyIUXXihrPgkhRCsKKqB99xmlD/wZfp57VA0EJkwm+6xLqUwRmlzXpRIVJbcLan4BruvWzVtqzrBUUcqmGy/Bq6necrxzAXm33kuFEWhSaHIch1CffmnL1awc0CUs7U52KuksXryYCy64gIkTJzZ3fdKqqanhhhtuYOjQofz+978nKyuLDRs2EAqFWqwOQgghkgVqqyi9/86k47HP38c/cm+MvfbHslI/Fed5XrP1KG0t0zYpu/WKhLAE4GxcT+XfbyV01a0/75W3c1zXRenWC61TV5zSDUnlGSeeScQfAKvxPfhE+7BTgSk7O5twONzcdWnQq6++Sn5+PhdddFH9sS5duqQ937KshD+giqIQDAbrf92QzeWNnbc76qhtl3ZLu3d3iqLgNwxcM45hGGkDzI7y+XxEX3ktbXnty0+ROXIc9uZNeVtKRRlu+aaUReb8H8mKRVH8TfsHd43mJ+/Wf1Bx1w1YSxfWHTR8ZBx7Gtq+k4jYTqt/xjriZx12Tbt3KjAdeuihzJgxg1GjRjVbRRrz3XffMWLECP7yl78wb9488vLyOPTQQ9OuBfXyyy/zwgsv1P9cVFTE9OnTd2g5hIKCgibXu73qqG2XdncsHaXdTk01zvo11Lz5ApvKSgnucyA5o8ejdyls8rU922bTpuQelvp7V5UT8PnI6JT+H7i7QnTd8gbLNdehsLDp7YcudL71XtzKcrx4DDUzGzU3H9UfIKMZrt5cOspnfVvN2e6dCkyKorBq1Sp++9vfMmrUKDIykj8WU6ZMaXLltlZSUsKMGTM46qijOPbYY1myZAmPPPIIhmFwwAEHJJ1/7LHHJtRhc8rcuHEjtt1wF6miKBQUFLB+/fp2sRZIc+qobZd2S7t3V0E8+PRdqh/7Z/2x2Hefo+Z1Iv9P91NhBJr0Hui6jn+v/Yh+9XHKcv/QUdQ6LpHi4p2+x87I6dQ1bZkSCOL4AhQ3Z52MYN1/HlBW3nzXbaKO9Fnf2o60W9f17epM2anA9NRTTwF14WPFihUpz2nuwOS6Lv369ePUU08F6nqMVq9ezXvvvZcyMBmGkfbR1u390Hie16E+YFvrqG2XdncsHaHdfjNK6VZhaTO3rJSaJ+/Hf/blRJuwKrVlWWQMG4OW3xln08bEQk0j49TzqHBb/n02fUGCBx5O9KN3ksoyfnk2EZ8fz27+uVNtVUf4rKfSnO3eqcB07733NsvNd0Rubi49evRIONajRw9Z80kIIdIwDAPzqw/Slke/+JD8My4kqjbtaa4q3UfeH++j+rF7iX39CbguRv9BZP36KmozcvHSPI2mqioBPHTHxtE0oqhNftx/s1oPcs64CK1rd2pf/y9epBY1N5/MU85DGb0PkQ4UlkTz2KnAtCPzgDzP48UXX+Tggw8mJydnZ24HwMCBA1m3bl3CsXXr1skWLUII0QDPjKcvdJ36ZQCawnEcynU/wXOvJDztEnBdHJ+falVP+wRcQIFA2QZq/vswkRVL0Aq6k3nyudiFvahtpo6QChf8hx9P3uSjMBQF04OI4cdqZFqGEKk037KnaXiex/PPP09ZWVmTrnPUUUexePFiXnrpJdavX89nn33G+++/z2GHHdZMNRVCiN2Lbdv4R41LW+4bOhJb9zXLvTzPI+LWLUhZYQSo9pS0YcnQNXxL5rHp6nOIz/wSZ9NGzLmzKLvhEvjyA/xa8z3ZFLdtKlUDo1tPqlRdwpLYabs8MDWX/v37c9VVV/H5559z5ZVX8uKLLzJt2jT222+/1q6aEEK0SZ7nYWflEtj3oORC3SDr3P8jorT810DYilP5rztSllU99k9CltnCNRKice1qZ8AxY8YwZkzqDRuFEEIkq/YUss++nPARx4FpgueC66H17kuV3rTVrndabXXaPdiwLdzSDdClR+pyIVpJuwpMQgghdoyiKKh4RL/6mMiM1/HMOHqvvmSd+39o3XrTPMtX7mCd1IYXsVRkSxHRBrWbITkhhBA7LstzqLz9t9S++UL9BHB71TLKbvwN/nUr0bQWXoEbIJyB1rVbyiIlGEbLzW/hCgnROAlMQgixm1IUBUrXYy1dkLK8+t9/I+S0Qh+TopBz7uVgbDPhXFXJOe//8DrYNh6ifZAhOSGEaGd8uk7IjEFtNYqu44QyqE3xCL+u65izZ6a9jrVyKZptge7f1VVO4FWUU/Pas3S5/X6iX32MuWQ+RvfehA44jOo3niPYuQClsHeHXGhRtF0SmIQQoh0JK6DM/IyyR+7Bq60BQO/Vl5wr/0BVZm5CaPI8Dy2vU9prKX4/NDKfaFdQfH4yjjiemrdexK0sRy/sib2phKqXniDjsGMgnClhSbQ5zRqYXNfl+++/54MPPuCaa64B6lZyvffee8nNzW3OWwkhRIejaRra8oWU3/unhOP2qmVsuu5C8u9+lPKtVu22bRtj2Ji6UOQmr4kUOngqMV8AWnptoowMot9+RuTDt5KKvJpqsi+/sW5PNiHakGYJTOvWrePDDz/kk08+oaKiAp8vcVxaVuMWQoimCzkWNU/8K2WZV1ON9dP36GMmJGwwXmv4yb32dsqn/w626n0y+g8meOxplDcSllRVbf59yOJxIh+/m7pozvd4sSj4w813PyGawU4Hpng8zpdffsmHH37IggV1Ewp79OjBsccey/77799sFRRCCFFH8zysVcvSllsLfkQfd0BCYDJdD6XfEDr981msOd/jVZZhDBmF17krFWik68rxfCFqLFhRFiUvZNA1w4fPieKmWb17R3i1NSl7vDZzykpRumXIsJxoU3Y4MC1atIgPP/yQL774glgsRiAQYOLEiXz22Wecc845DBkyZFfUUwghOjxXAb1rN+ziNSnL9d79MVMEmrjrEtd8GGP3o3Pnzqxfv/7nBSvThKVAJne8v5SvVmzZ0io7aPDXY/eki2HhNHEITwmGQFHS7mOnZudKWBJtznYHpjfeeIMPPviAtWvXAjBo0CAOOugg9tlnH2zb5rPPPttllRRCCAER3U/GL8+h4m+3JBcaPnx7TaTWSr9MwOaep4bCiObz8/jMdQlhCaAyanHpiz/x2Kkj0anZuQb8zPQHCew1kdg3nyaVGX3644QymnR9IXaF7Q5MTzzxBACjR49m2rRpFBQU1Jel22BRCCFE87FtG2XYaDJO+BU1Lz1VP6ylZmWT+7vp1PgC4DRtq5OYZ/DqnOKUZTVxm+VlUQZmqU3aUiXiQc55V+JFI8R/2rLsgdGnPzm/m06FqkMzbtmi63VfdZqmJQxXCrEjtjsw9enThxUrVvD9999TWVnJQQcdxIQJEwiFQruyfkIIIbZS5SoEjziRTgdPxS3dgOIPQHYuNboPu4lhCcByPeJ2+uusq4oxOCfYpMDkeR7Vuo/sS2+AqnKc0hLU3HyU3HyqdT9uM4UaQ1PJMGOYMz+lbPUKggOHoQ/ak2ojIMFJ7LDtDkzTp09nxYoVvP/++3z++ec8/PDDPP7444wbN45x48btyjoKIYTYStT1iGo+lIJeW4bXmiEsAfg0yAsZlEVSD+0N6JyB45hNuoemaWTVVFD6+wtxqytR/AE8M47i85H3h3txO3fDbuLIha6phDesZdNNl9ZvCcObz6Nm5ZD/p/uoCGa2zsbDot3aoa1R+vTpwznnnMMDDzzAb37zG/bYYw8+/fRT7rrrLgC+/vprKioqdkU9hRBCbGNXTIwOeCbn7dMnZVlRfoiuYb3J9w07FhV33YBbXQmAF4+B5+HF45T/6Roy7HiTrg+QYZuU/+maLWHpZ25VBRV/uYmwJ1NJxI7ZqWUFDMNg4sSJTJw4kZKSEj744AM+/vhj3nnnHWbMmMGYMWO48sorm7uuQgghdjHbsti3VyZXHNSfh75YQXXcRgEm9M3jigP7oVu1NLVfRovWYq1cmrLMrSyHygrISb9C+XYpK60PZNuyli1Cj0VkrSexQ5q8cGWXLl04+eST+eUvf8mPP/7IBx98wMyZ6fcuEkII0caZESb3DjGhz0iilouhKwQVF8wa3Gbo1fIamT/kxaMoirLTPVmKouDGIg3fw7KgZbfQE+3cdg/J1dTUcNddd6UNQ4qi4DgOruty//33N1sFhRBCtDzbMtHNGjK9CAGrFs+MNt8QYDgTJZimd0dVUTt1bdK9PM9D69q9bq2nFJSMTJClC8QO2u7A9MEHH7By5UpGjhyZ9pyRI0eyevVq3n039ZL3QgghRMTwkzXtopRl4aknEzOa3vUT9wUIH3ViyrKsaZcQaYZ7iI5luwPT559/zuTJk9G09Dtba5rG5MmT+e6775qlckIIIZqHYRjYRogNVTEwAqjqDj3z06xMx8EbO5Hc301H79EbAK1zV7Iv+T2+Kb8k6ja9JyvieviOOY3si65F69QVAL1nEbnX34U3anzKFdGFaMh2z2EqLi6mX79+jZ5XVFTEiy++2KRKCSGEaD5eIJO3F5fyzMy1VEQtRvfM4aKJReTpNp7dtCUCdlatp6APGEbmTX9H9VxcFKK+AFYDK5XvqCpPQd9rf7JHjcdv6MQsm4jukzWYxE7Z7sDkOE6DvUubyUqqQgjRdni+MH/63xK+WVlef+ybleXMXFXBA78cQYFPa7XdGmzbpgoVlJ97u5oxLCXcQ9Eo7FxAdXFxoxPOhUhnu/tkc3NzWbMm9YaPW1uzZg05OTlNqZMQQohmoCgKm2JOQljazPE8/vLRUmxN5vIIsT22OzANGTKE9957r8HeI9u2ee+99xg6dGizVE4IIcTO03Wdb1ZWpC2ft74ay2u9uUxCtCfb/SflqKOOYu3atdx1112UlZUllZeVlfHnP/+ZdevWMWXKlGatpBBCiB3neR4Z/vQzL3RVQU395L0QYhvbPYepd+/enHPOOfz73//mkksuoW/fvnTp0gWAkpISli1bhud5nHvuufTq1WuXVVgIIcT2sW2bvXvloACpnjs7ZGAX/NjI82JCNG6HVvo++OCD6dWrFy+99BJz585l8eLFAPh8PkaOHMkxxxzDgAEDdklFhRBC7LgAFr89eA/u+N/ihOPdswOcu08vnHh1K9VMiPZlh7dGGTBgANdeey2u61JdXfcHLTMzs1XX9BBCCJGGHWefHiGe/tUY3l2wkZKaOPv3zWdglxBGA/vCKUaAGDoVUZuAoZJpKBhOLO0TdZquE1f9VMVdbNcjN6jjd03cVlq2QIjmttN7yamqSnZ2dnPWRQghxK5gxQkT55Q9c8nPz6ekpAQ7XpN+E11/Bg9/s5Y35hazeQ3JXrlB7vjFULK0aFJo0nSDlRGV69+cTXmkbmkAv65ywYQiJhVlgRXddW0TooVIt5AQQnQQplnX29PQuku64ePVeSW8NmdLWAJYVR7lNy/OxtSCSa+JKH4uf2lLWAKI2y5//3gpSyut7VrDT4i2TgKTEEKIejHFx7PfrwUg7NMY0CWDwqwAAJtqTVZXxhOmYPh8Pt5buBHLSb2dyUNfrGz2tZ50XUfXt3+AZHN9lTSb8TYHTdPQdX2X3kO0rp0ekhNCCLH7MR0PRVG49pCBZAV0lmysJTdkUJgV4IlvV7GyPEL/zCCu+/OAnqKyaGNt2uutrojioDTPl40RIOJqfL+6EkWB0T1yCCo2WLGUp2uaRlwLsqgsyvurV9EvP0T3TB+6HdlS/6ZWSVXJsOPYC2fjlpdiDNwTN68z1ah4XtP3xBNthwQmIYQQ9Xwa/HHKUP756VIWldTUH/frKjcdMZiumQaOE9/yAs9lSNcMPllSmvJ6RXkhtPSzpbab5wvx4pyNPPHt6oTjZ+7dk2OHdgYzknBc0zTKXD+XPDeLqtiWBZe7Zwf4+3HD8Fk1TQ40hqoSKl7Jpj9cgWdueU+MPYaS89s/UY70Nu1OZEhOCCFEvaAG78xfnxCWoG5O0h/emU92wJ8QNEzTZPKATvj11F8nv963N5odT1m2vVRVZWWllRSWAB79ZjVrqu2kJ7VNLcCVr8xJCEsAaytj3PbeIjwj0KQ6AWTYccq3CUsA1uK5RJ5/hIAuc7d2JxKYhBBC1Is4CjMWlKQsi1kuSzZWJ4WToBvn3hOG0zVzy1ylsE/j94cMoFem1uDwl8/nw/AH8PsbmOek+3lqZvq9TJ/+fg2Knvj6sqjDxprUSxrMWltJ1E0fZjRNI+Dz4ff70s5J0jQNe/H8pLC0WeSDtwiYqYcKRfskQ3JCCCHqWY6H7aYfqtpYY6LmBRJCkGNbdPN73HfCntRYHo7rkeXX8HsmTrr5RbpOTA3yzuJS5hRX0Tc/zJGDuxAiee0m24OySPr1nDbVWtge9QNgiqJQHU+/7ylA3HbY9nk/RVHI9mychXOJf/wOSjBM9uHHYud3ocZTks51yjamv4Flgm2Dz2iwHqL9kMAkhBCinl+Drpl+NlSn7jkZUpCROIfpZ7Zto2GTDXXJxSTtliuaplFiGVz0/ExiVl3w+nTpJp78bjV3H70n/bIMXHvLEgU+xWFsz1yWpJlcvlevHAxcNkckz/Pqn+xLJWhohH0abJPBcnCouO0q7JVL649FPnyL8FEnknHM6dRsNSfJcRx8A9JvNK917YZr+NKWi/ZHhuSEEELU87txLtmvb8qyPQszyQ9oTZ4sbal+bnxrQX1Y2sxxPa5/az6mus3wnOsyZc9CgkbyMFrI0Dh8SAGem9ijFFQcDhnYOeX9p+3di+A2acmn68TeezUhLG1W++bz6BWlCcNznufhduqC0X9wyntknfUbIkbzLqcgWpcEJiGEEPVs22ZYFz+3HTWYgp97aXyayjHDCrn1yEGoVqSRKzSuxoY1FalX/66J25RFE8OPrRg8+/1q/viLoQzvvmWHiZHds/njL4by3MzVuNo2vTlWlEsm9mHa3r0I/Ry0ckMGV03qzxEDcrHNxMAUtOJE3ns1bZ1j/3sjaZ5VNRrZ195O6JCjQa8betO6FJJ77e3Y/Ydi2w0PC4r2RYbkhBBCJFCsGKM7G/zr+KGYLuiqQgAbJ1ZNc6ws5DQwRwrAtF3YqjPJBRaW1PDF8jKOGd6N08f2BGDBhhpufWc+BVkBUnZ6xao5aUgOU4d2AVVHcW0CXhzLTBX6vLQTuAHcaG1SD4PneZSjEjj11+Sf8CtwbDzDT63uk7C0G5LAJIQQIollWWhY9ROj02+msuMy/SrZAZ3KWHKo0FWFrpk+2KoHyI/NIQM7889Pl/PvL1ckveZXe/XCwMZKKgHbMjEUi8LCQoqLi7HSDCeauo/AuAOIfvhWyvLAgUdQY6aeeB6zHWKqAerPE7wlLO2WZEhOCCFEiwq4JldO2iNl2Xn79sHvJUYfy7KYvEenhGULNivICrB/vzwsK1Vc2n4x1yPjpDNRQuGkMmPAUJSeRc22Orhon6SHSQghRJMpioJi+DHR8QADB9WOpwwZtm0xorOf+08awYNfrGRJaQ3ds4OcM743e+QZuGby/CbDquX+k4azoCSCoavgge26DOwcQjdr064l7g+GqPUMVmyqRfNlkqFaRKPJ1/c8j0p/mE53P0rtS08S+/pjlECQ0JHH4594CBWopB73Ex2FBCYhhBBNoqoqcT3Mw1+uZMaijTiux4ju2fzfgf3I162EJQLq2XG6+1VuOrgIGxVNcdEdEydFWIK6QBazPV6fU8yXy8tQFJjQN58+E4sw0m14G8zm5fkl/Pf7NVRGLQqyApyzTx/27pEFsaqk0x3HoUz3Ezj112SfMA0UhajhJ2Lb0Cyzt0R7JkNyQgghmsQywlz8wmzeWVBSP6H7x7WVnPffH6hy06+W7bouWFF0qxbFjOI46WdKxbQQ5z7zA18sL8MDXK9u7abz/juLmBZKfkEggwe+WMEDny+nMloX2NZXxfjjuwuYsXgTvkCK12y+l+1QpepUKRqWzEcSP5PAJIQQYqfpus6sdVUpF7q0HI+Hv1qJ0sT1iHTDz6tz1lNrJgeq6rjNW/M2oG+zSGTUUXl73oaU13vkqxXUeDLAInaMBCYhhBBJNMOP5cuggiBRPQN8wZQ9Raqm8/GSTWmv892qCkyvaZvQmqh8uaIsbfkXK8qxtlqHQFVV1lRE0w6i1ZoONfHmfO5PdAQSsYUQQiTyh3ltwSYe/3ZV/Wrco7pnc92hA/DbtQkTuRU88sPp90vLDhqoStPm/2gKZAXS3yMnaKApXv3Eb9d1yWhoM1/Ar6sgo21iB0gPkxBCiHq64ePDZRU8+MWKhK1LflhbyeUvz8HWE+f+mKbJ0XsWpr3eKaO743PTb5y7PVTH5PQxPRq8B9ts2NspbJAbSh2yhhRkEpLuArGDJDAJIYSoF1MMHvlqZcqyNRVRimvMpD3VcgyX3+yfvP/chKI8JhTlNHnVa8dx6Jfr49jhycHsxJHd6JNtJC1fEPRi3Hn0sPptUTbrFPZx4xGDMezUT+MJkY5kbCGEEPVMh5QrcG+2pLSW3r2CiSHIinFwUSYTisbyxfIyIpbDPn1yyQ+oKLGaZqpYhGmjCzhueCGfLS9DBfYtyiNL9yDFVie2adI9oPDYGWP4cW0VK8siDCnMZEDnDIJORJ5+EztMApMQQoh6hqYQMNSE4bit9cgJpl7x2o4TJM7hfcMoioJlxfHM5l27SLGiZAFH75EJgGVFSLkfys8sM46POBO7+zl6eDc2btyIFats6CVCpCVDckIIIeoFPIvjR3RPWZYdNOidE2hwixDLsjBNE28XroptWdYObYVi/rwHnGyIK5pCApMQQoh6jhXnxBFdmTygc8LxLhl+7j1+OH6n4bk/Pp8Pny/9YpXb0jQNv9+PrsuAh2jb5BMqhBAigRKr4dJ9e3DuPr3YUBUnM2CQG9DwOQ2sxm0EqHE0PlpQSsR02K9fPgVhHcWsTXm6ruvE1AA/bqhh7voK+uaFGdMzm4AbxZGeINEGSWASQgiRzIoSAvqGFTzPBhPSLvVoBHlzUTkPfrGi/tBTM9ewV69crju4H0o8ceK3qqpUen4u/O+P9duWAAQMlXuPH06Bz2twmxQhWoMMyQkhhEirsblIiqJQZpIQljb7dlU5Hy8rxzAS10Ny9AA3vr0gISwBxCyXa16bi6kFm1xvIZqbBCYhhBA7zefz8epP69OWP/fDWmIkBqZaW2HJxtRDdWURi/JY+t4lXddRfEFUX3C75j0pioLm81MVs5L2m0vHMAwUXwDVF0TTmrati9h9yJCcEEKIneahUBFL/8RaVczGI3ECuOWkf8oOIGrZKAEloXdLURQcX5hvi6t57ac1ABwzrIBhhRloZm3KnjBF91Nmazz71VpWly9jSEEmxw0vJEwc106us6qqWEaYj5eX8b9FpQR0lV+O6ka/vDBKPHXAEx2HBCYhhBA7zXVsJvXvxEeLS1OWj++Ti09x2DoiZfo1QoZGxEruSVIV6JoRwLOqE447vjDXvbWQeeu3HP9+TQVDCzO57YiBqNvOk9J9fF8S56a3F9Qfm1Ncxcuz1/GP44fTM6gnLTNgGWEueuEn1lfF6o99s7KcyQM6c8mEnqgpFsgUHYcMyQkhhNhptm2zZ0EG3XOS5x35dZWzxvXCs+IJxwOeyQUTi1Je76SR3fFvs7Skrut8u7oqISxtNre4mlnrqpOG50zVx59mLEo633I8bn13IaaauDmv7vPz3Kx1CWFps/cXbWRDrbPdSyWI3ZP0MAkhhGgSw47wj+P25PMV5QQNHU1VqIiY7N07h0xM3G2Gy2zLZP8+meRPGcL9ny1ndUWUzhk+zhrfmwm9svC2Gf6yFINXfkq9vx3Ayz+tZ3RhX6Cux0hRFNZXm8Rtl1+O6s7U4d0wtLrVyx/+YhmfLN1Ere2RudU1Yp7O2/M2pL3HG3M38JvxBcRiyYFKdAwSmIQQQjSJ53koCmyqMXl1zkrilsvEfvnsW5QHbuqn7BTPo2uGj3P37UNA1zAdl5ygVnetFOenuUxdmevBNuWu5/H0tL1YWFLN9W/MZV1ljH6dwpw/oS8nj+nJtlOeFKXheziut0tXLxdtnwQmIYQQTeL4wlz1yjyWbtrSM/TeghI+W7qJR04bRcBJnF+kGQZfF0e59d2FSdc6Z3xvjh6UkzCMZ3gWRw3pyoINyUNyAL8Y2hUDu34gz/M8eueHefb7tTz13er68+atr+ayF3/k+sMGMbEoh3j1ljlJPmwOHtiZl2cXp7zHUUO77tB2LGL3I3OYhBBC7DRN01hQEkkIS5tFLIfHv12N6kucLxRX/Pzto6Upr/foN6swlcRlCGzbZmJRLr3zQknn980Ps3ev7KQwE7Vcnpm5Oul8gH98spSabTYGdsw4p43pQV7ISDp/XO9cumcaDe6hJ3Z/0sMkhBBip2m6wXsL16Yt/2zpJs7eqwc6W3qMquMO1fHU2584rseGapMeiRkL3arlb8cM5eNlZbwxdz0KClP37MrEojw0syZhRE7TNNZURNMOsVVGLarjNnnbHA84ER46eSRvzdvAB4tLCRoavxzVnRGFGUmrlYuOp10GppdffplnnnmGI488kjPPPLO1qyOEEB2WgkfYl35xx4ChoWwzwUjXGh7c8OnJ5a7rosarObRPiAP6DEChbqjOjldvO30Jx3EIGMk9RQl1UBXYpsPIcRx0p5rjBucwZVD+VveQsCTa4ZDckiVL+N///kfv3r1buypCCNHhmabJMcMK0pafMKIbgW2WCQgbKr1zk4fXALICOnmh9CtyW5aFakVRrGjSOkpb65rhJ8Ofuk+gX6cwGf70Ic82ze26h+hY2lVgisVi/OMf/+D8888nHA63dnWEEKLD8zyPTgGFk0d3TyobUpDJIQM7pZwsfflB/ZN6pgxN4erJA5qlXiHV5tajhtT1JG0lw69zw+GDCbqyPIDYMe1qSO7hhx9m1KhRDB8+nJdeeqnBcy3LSvhDqigKwWCw/tcN2VzeERcp66htl3ZLu3dnhmEQVQJUmQ7Fq8vJC2URVB3cWPNs96HYMU4e0ZVDB3XhjbkbiJgOhw3qTJ/cQN22Jdu8zxUxm39/uZx//XIUjuNREbPI9OsEDY2HvlhOblCnd1BNeoxf1Q3iio/Kn/eayw5oBDwTJ8U2J3YswsD8DJ6athfvL9zIsk21jOiezT5F+WQoccyY2SF+/zvaZ32zXdFuxWsnC0t8/vnnvPTSS9x+++34fD5uvvlm+vTpk3YO03PPPccLL7xQ/3NRURHTp09vodoKIUTb4Loui0pq+d3rc1hTEQVAUxVOGNmd0/fqRUFWoNnvWbcuU/ovqpWballeFuGV2ev4fNmm+uPDu2Vx7r5F5IcNBnTJSnhNxLT5cnkZt727gJqfJ4xn+nVuOGIQ43vnEfQl//t/XWWU61+fi6Gp5IV9rK2IUpDl56rJA+iS2fztFru3dtHDVFpayqOPPsp1112Hz7d9u00fe+yxTJkypf7nzX94N27c2OiYtKIoFBQUsH79+g63UFlHbbu0W9q9uzJ9WVzy/A9Uxrb8vee4Hs9+v4YumX6mDMgmHo22aJ18gUzemrs+ISwBzF5XxX2fLuNPUwZTXLxlPSRFUShzA/zutTkJE7yr4za/fWUOj50+mlwllvB76RghfvvmQhaVJE7Ynr+hGr+ucsn47rjW7j8s15E+61vbkXbruk7nzp0bvWa7CEzLli2jsrKSa6+9tv6Y67rMnz+fd955h6effhpVTZyOZRgGRpqnJLb3Q+N5HXdl147adml3x7K7t9swDH7cUJ0Qlrb21LerOKD/KIxmfA80TUNRlAb/YRq3XT5esjFl2fwN1dSaDplb1Uk1/Dzx+eqkp+GgboHvZ2au5eLxhbjmlgBUbXn1YSk/7CMnaFBSHac6bjNjwUbOHtcL/278e7+t3f2znk5ztrtdBKZhw4Zx1113JRy777776NatG0cffXRSWBJCCFEXmJZsrExbXhaxcFxo+AH87aT7MBU/c9ZXETUdhnfPJkN3Uczk3qta02lwG5LyiElWUKn/orM8heVlkbTnL9tUi+0p9VuqKIpCaY1JUX6IX0/oS23cZn11jN65IWzX495PlhK1HPxpryhEsnYRmILBIL169Uo45vf7yczMTDouhBCijmVZDOiSmbY8P+xDb4Y5sYru5+viKLfPmJ0QhI4Y3JXz9+mZtOhj2KehNrB3W27Ih+dumchtKB79OoVZWpp6knr/TmF0xcP5+WfP8+iWHeCyA/tz81vzqYhuuVbPnCA3HzG47gk9c6eaKzoo6ZoRQojdlGVZDOqaQU4wdR/SGXv1Iqw2fX+0Klfnj+8tSgpAb8/fwHdrqtH1xH+bBxSbQwZ2SXmtPQuzyNATL+RacU4b2yPlpryqAqeM6YG71d5zAIam8qd3FyaEJYDVFVH+89VKfFrHempMNF27DUw333yzrPIthBCNCLlR7j1xZMI+bIamcNrYnhzUP7/BCd+6ruP3B/D70w9e+f1+Xp+zPm35E9+txlQSH9bxzBgXTujN5AGdE0LQmJ45/OGIgahWYp08zyNXt7lj6lCyA1vCV3bQYPrUoWRpVtI8lYqoSUlNYoja7Ic1FUSsjjefRzRNuxiSE0IIsXNM0yTP8PjrsXtSHXeJOy65QZ2w6mJFU89v0jSNuBbk6zVVfLFiPYWZAaYM7UqGaoOdGEI8FNZXpw4mAGW1ZtI6TABKrJrLJvTg1/v0pjpuE/ZphDQPxaxJOUnXs0yG5hr855SRVMXrBt+y/Bp+N45jJY6tKYpCVZqJ7pvFbYdgg2cIkUgCkxBC7OYsy0KzLHIVhcIehRQXF2OleXJIURQiaogLn/uRTbVbgsgz36/hukMHMK4wgGdvOe46Nvv0yeXjJaUprze8WxY6buon3MwoASCgAU7dfw31+zi2hY5F3ub8ZVI/bynhup5HYQPrSwUMVeYwiR3WbofkhBBC7AKGn79+tDQhLG12x4zFxNXE4Tnbttm7Vw754eQ18jRF4df79oFWWO8oqDpMGpB6bZ1f7dWLgNf0uVuiY5HAJIQQzUw3fDhGENcIpV0Prq2KeTpfrShLWeZ4HnOKq9C0xD3gfHaE+04cwT598urnJPXrFOafJw4nV7dbZ/0fM8pl+/Xh9LE9CBp19c0OGvzfgf04cmAejpV+GFGIVGRITgghmomqqlhGmLcWbuSd+SXoqsLUPQvYv28eWpq5OW2N63kNDotFTQdFSQyBjuMQcGv43UG9iHl9cT2PgOqhO3EcO9WgWQuJVXPynvkcO6wAT9VQXIeAZ2Kb6dd0EiIdCUxCCNFMLCPMJS/+xLrKLUNQf/1oKa/PWc+fpw5G3WY9orYoqEHf/DDLNqVe82h49xxsO7kdnueBGUtYDLIVo1I9x4pjKCaFhXVzt+x2EFpF2yRDckII0Qx0w2DGwo0JYWmzJaW1zFqXvB5RW6QqChdMLEJNsUzRQQM6EzDka0N0TPLJF0KIZhBH550FJWnLX5+7AVdt+4EpYnu8MWc9dx0znHF98gj7NHrlBrlk/37sW5TPdyvL0wY/1efH9mVg+TLAF2z2basURUH1BbB9Gdi+DBRfoH5j9XQMw8A2QqyrjOIYoXYRWkXbJJ8cIYRoBgqgpeqW+ZneQFlbogCfLC3lx7UVHLVnIYcM7EJ13OadeetZWFLDdYcOSJqLpaoqphHmP1+v4r0FJViOx5ieOVx2QF9ydRO3gY14t5emadQoQf71yXI+W7YJgP37deLCiX0Iu1EcJ8UAoD+Dd5eW8fi3CyiPWHTPDnDBhCKGdQ2m3ONOiIZID5MQQjQDn2dxzJ6FacuPG16I4rT9R9n9qsvoHtlUxmye/m41t727gL9/tISFJTUowIhuWUnhxDJCXPLCT7w5dwOWUxemZq6u4JxnfqDa8zfaC7Q9YlqI8/47i0+WbsL16vah+2hJKef9dxYxLZT8AiPAI9+u5e8fL6M8Uve+r62MccNb8/l0RTWakbwMghANkcAkhBDNwLZtJhTlMqBLRlLZXr1yGNg5mLoXpI1R7ThXTupPhj95AOKi/Yrwb7N+ka7rzC6uprgqee6W5Xj8+6tVKHr6rVW2h274eW3ueqrjyT1VVTGbdxaUYGwTgGLovPpTccrr3f/5cuKKBCaxY2RITgjRbmmGn5hiUGs6BHSNgOqgWLFWe3xfM2uYPmUQs4treGPuBlRV4bjhBQzoFGoXT8gBuK5Lthbj0VNHMWPhRr5ZVUHnDB8njexG5wBJi1Cqus5HS9LvJffNynLMfXs16cvGROXL5eVpyz9bVsaUgVvWgFIUheLKWNrlEWpNhxrTIacJdRIdjwQmIUT75A/zyvxSnvxuNTHLBWB8n1yumdQfw6rFdd0Wr5LneajxGsZ20Rld0LvuC9yxcNpJWNrMcRwMp5qpA7I4amAumuLhmHG8FBvWKp5HXjD94pxZQQNVaVqA1RSPzED6r6vsgI6meGz+Hfc8j5Cv4a83n6a2jXUPRLshQ3JCiHZHN3y8u6iMh79cWR+WAL5aUc7Vr83D1lt3W1XbtnHNGI4ZaxfDcOnYlolnRrHj6XvtTNNk6rD0c7dOHtUdn9u0TdtUx+LUMd3Tlp8ypgfYiffID+nkpAlyQwoyCWmyHpPYMRKYhBDtTkzx8fi3q1OWLS2tZVPUaZaJxltr7kfkdxee55FruFw4sSipbHzvXPbrm4vdxKfkHMehf26AX+xZAEDYp9VtngscO7yQPtlGUo+i34ly19FDk9aNyg/7uPGwgah2y+9vJ9o3GZITQrQ7MdulJsUE4M2Wl0UoKDCa3LujKAquEaLS9NhQGadLRpAcv4pmR1tlyK/NsmIc1j+b/fqO4fNlZdSaDhP75tEpqKHEqpvlFopZy3l79+D0vXqxfFPd1iZ980MEVBcvljzk6TgOhQGNJ88Yy6y1VazYVMuQwiwGdgkTsCPY8vsndpAEJiFEu+PXVAxNqX+EfVuFWX5ct2m9GoqiYPky+P0b81lYsuULuW9+mDunDsFvt848qbZKsWKEgaP6Z9S9d1Ycz2zGYS8jwGcrK/jLh0ux3brrGprCVZP2YHz3UPJkdFWlBh9XvvQTuqqQH/bx/qKNdM7wccvhA9Gc9rG3n2g7pI9ZCNHuBBSLKUNTz5vJD/sozPQ3+cvQM4Lc9u6ihLAEsGxTLde9OR9HDzTp+rsry7IwTbNZw4iiKGyIwp3vL6kPS1C3bMHtMxZRGleShmBtPcjVr85lTUWUFWURZq6uoLgqxux1Vdz90TIwmrbUgeh4JDAJIdodx4wzba/uTCjKTzjeNdPPPccNw+80fRXniKPww9rKlGULS2qotVt35W5FUfD5fBhG+ifUmkpVVXw+X6tvJ6IYfp78LvWcNYCnZ65B3SYAVcRdVpWn/hx8vmwTUU8GWMSOkU+MEKJdUmLVXHNgb2r268O6yhi5QYO8oI7PSbNNxg6KWg1fo9Z0yGiFv0EVRcExQqyvtfliYRlZAZ39+uYTxAI73iz3UFUV2wixtDzGzNXlFGT5Gd8rB79n4tlNe+JtZ9iewtoUmxpvtqYiiu0pCeswbV7dOxUPiFkO0kcodoQEJiFE+2VGyAAGZip4XgzM5ltaJ8Ovoyp1W3Ckkh3QoelbpO0w15/BTW8vZPa6qvpj//x0OVdN6s9+vcJgNS00KYqCZYS57KU5rK7Y0kOjKQrTpw5lQI6BZ7fsFi+G4jKkIJOlpbUpy4cWZmEobv1vh+d5dM1IP+RmaErdU3Ytn/1EOyZDckKIdm9XTN4NKDaHD+6asuzA/p0Iqi2/vpJh+HhzXklCWNrsrg+WUOM0/d/Aiu7nn5+tSAhLAI7n8bs35mKqrTD3x7E5YVSPlJsb66rCsSO64W4T4sK6x9ieOSkvd9yIbgRo+/v6ibZFApMQQqTgmTF+vU9Pjh3eDf3nL2pNUThySFcuP6AIrxV2u48pBs/PWpe2fMbCjfh8qfdIUxQFzeenJm6jNzDvKYbOR4tLU5ZZjseCkho0TduxijfAHwpDIBNfKCPt2lmuZjBj/gZuPWoI3bK3DKR1zwnyh6OG8O78DaAntlu1olx/6B5MHtCZzTnLp6mcMqYHp4wsxDGbZ/hSdBwyJCeEEGkosRrOHtOFU0d3I2o5BAyNoGLjNtPaQjtRI6pj6ccBN0XM1KHDF2Rdrct/v1pDedRiQlEek/foRMCNJi0qabseTgM9dhVRG0Vp+leH7g9S5eo88e1aFm2spWdOkF+O7kGOz8OLJw69OZ7C1yvL+WJ5Gb/auzd5IaO+vQ9+vpyAoXHC0E5s3XLP81Bi1Vw2oTvn79sb21PwKR5+xcJtZ1vViLZBApMQQjTANWP4AB+ABa258pKGw+ieOXy7KvVGtPv3y09eVdsI8NLcUh77ZstTZj+ureSpmWt48KQRBJXahCFNvwbdswNpJ1nvWZiJ4zRtlWy/38/Sao/LXvyufpmA2WsreWvuem45aghjuwax4lt68Hw47FuUy2PfrOaOGQuTrnfuPr0xcFJOKfPMGAFFobCwkOLiYlxZe0nsJBmSE0KIdkK141yyX1HKuTy980L0zw8mPSFY62oJYWmzyqjFPz9bAXrinCS/G+f/Duyf8v57984lx6c0ec5YLX7+8M78hDWVoO7ptTveW0gtiUOGtmXyi6EFZPqT/42fHdA5fFAXbEtmcItdS3qYhBCilem6jqn4cBUFHRfVjqdcRdx1XfJ0i0dOHcmGGotOYT+261FSHWNI1wx0syahB8wwDL5alro3CuCzZaVcun+fhHhi2zZ75Pq55/hh/P3jZSwtrSXDr3PSyG4cvWdXaGA40jB8xDDwFPDj4Jipe6KqTYcN1annEEUsh401Jt23mWYVcCI8dPJI7v98BZ8srZtjdWD/zpw/oTcBJ5L26Uhd17FUHyXVMTwjiGrHmn2FdsMwfm63goELVvrNikX7JYFJCCFakecLMXNDhEe+XsaGqjgDu2RwwcQiCgJeynWVVE1DR+fteWv4ekUZYZ/O0cMLGVqQlfL6De2Z5nqQ8nvditM3rHHXLwbioKICfkzsNGFJURQcfwYvzSvhldnFxGyX/frmc9a4noTc5HWxGssSqYbNHMchpNRyxX49+M1+fQDwYeNZtWnnXHm+ED9ujPHvrxZSXBmjf+cwF04ooluIZluzCn8GH6+q5KnvllAeMRneLZsLJvQhV7dbZc0qsevIkJwQQrQWI8Bzs0u48a0FLN8UIWI5/LC2kvOfncWCcjtphW1VValwDM56aiYfLNpIrelQUhPnoS9W8Ps35mL5MhLOtyyL8X3y0t5+r145+JTUfTOO46CaEQyzBs2swTbTf/k7vjDXvDaPh79cSWmtSU3c5u35Gzjr6R+IqMGk87MCev3E7W35dZWumamXLvA8D8+MoZs1db1pZvqeHMUI8Oq8Un7/xjyWltYSsRxmr6viwud/ZPZGE01v+grpni/I3z9dwfT/LWZNRZRa0+HLFWWc/fQPFMfUZn2aULQ+CUxCCNFKYug8M3NNyrK7PlhMfJs1jzxfiPs/XUbcTu41mre+mlXlsaSQlaV7TBmavJ5UwFC57IB+qE3sadE0jYUbIyzemLyoZK3p8Pi3a9B8ie0IeTF+e/BAUi0icPF+fQkqTV8RNI7O4ynmbgHc/eGSZllPqspS+CDFEgyO53H3h0uwNdmvbncigUkIIVqBqqqsKIuQbnRqY41JrZVYaqHx5YqytNd8f1EJgUDihh+KFeXcvXtwxy+GMKxbFj1zghw/ohuPnzaaHCX1XKkdoesG7y7YmLb806WlxLeZ/WHG4wzq5OM/p43hgP6d6JETZO/eufzrpJEcUJSDE4ukvJaqqni+IDE9TEwPgy+EqiZ/jSmKwrqqeNqhuoqoRY3Z1HbrfL869V6DAItKaoi7rbvfoGheModJCCFaged5BBrZ1FZXlW32evEIGBo18dQ9MCFf6iEgxaxlz1yN2w7rhz8YwonWYps1zbSNjJf2vgABQ0NJFQvjETqpKlcf2Ie4q+BTQTEjOPHUE8U13WCTYzD97cXMW183l2pYYRZXT+5PnmbiOFveE8/z8OsN9wdoqtLkNSIaareqgJpmIU7RPkkPkxBCtALP8+iR7SdgpP5reFDXTIJaYtAIqzZHDilIe83DBxdQU5N6UUbHcVCsGPlhP04T95vbmmmaHDMsfZ2OHV6I30u9DYnrujjRGvR4NW60usFNk2sVP7/+76z6sATwU3EV5z/7IxEteZ5U57BRt19cCkX5IcJN7C6wbZuR3bNIscIDABP7dsLfDEOLou2QwCSEEK3E58a59cjBSV+6mX6d6w8dgGYn9rbEIxFOGdODPnmhpGudNrYnuf6W/yvd8zw6BxVOHNktqWxQ1wwOG9QZ22ravm2a4efFH4tTzt2KWg6v/lSMbiTOF/K5cf44ZUjSmlVhn8bNhw9Cb+LimwABz+S3Bw9IOt45w8dv9i9q8kbIom2RITkhhGglrm0xMMfH078ay7sLSlhRFmVMj2z26ZOL34ngOMkBQY9X8ZdjhzF3fTUfLt5Ipl9n6rBC8gMqSrx1tmxRzCi/GlPI8SO7s6nWwnZdOoX9ZPoUiFWnnae1vSxUvl9TkbZ85upKTtyzc8Ikcte26Jfp46kzxjBj4UaWbYowonsWE4vyCLhRHLvpA5KebTK+e5AnzxjDm/M2sL46zsSiPEZ2z8JvRxrcYka0PxKYhBCiFXm2SRCTk4bkoCh5uK6DZVannV/kui5arJKxXXyM79EHTYHa2lrceOtt2qLqButqHK5/8yc21tQtPxAwVH6zf18m9swAq2m9Obri0SnsY0mKJ/EAOoV96IqX9J65P7+3Jw7JITe3L2VlmzAbeG93ihUnkzjTRuQDCo5jY8eb+R6iTZAhOSGEaANM0yQej2Ft5/CVaZpEaqqprq5u9pWrd1RU8fObF2bXhyWAmOXy5/eXsKLKafp6RLbJaWN7pi0+bUyPBheJNE0TVVW2+73dGfF4nHg8lryXn9htSGASQrQpqqomrSXUHmma1uYWLtwV76vP5+N/i0oxUwwfAjz45UoczZf29YqioOt6yuUBNnNdl16ZOmeN65X4WuC8fXpTGFZbPTSK3V/7/1tJCLFbUHWDuOpn8cYIFbE4QwsyyTLqHjVvV4wAUU/np+IqdFVhaEEmfqxWnQCs+/1ECDCnpIZNK1exZ7cs8gI6Sqyq6RdXVBaUpH4yD2BlWQQbNenLRlVVbD3I+lqbFZsidMsO0DM7hM9J3koFADPCMYPzOGxQF2avq0JRYHi3LALYYEWb3g4hGiGBSQjR6lTdYHGlx7Wvf5/QU7FvUR7XTOqH0sCGr22J5wvxwk8lPPndltW7FeCS/fsyuW8WShPn8uwM3R9gcYXLb1/7NuEps9E9crjx8EEo0Yqm3cBzGdQlg0+WJK94DdArN4jmJfb+KIpCXA9z+ctzWFOxJezkh3384/hhZGmx1KHJihEE9i2s29bEsdIHNSGamwzJCSFaXUwJcPVrc5OGdb5YXsYb80owjPRDOm2FpmksrzATwhKAB/zjk2WUxuqCQkuL4OfqV39KeiT/+zUVPPvDGnzB5CUKdoRpmkwa2AWflvrr5KxxfQgbie32jCC3vrcoISwBbKo1ufKVuZha4mrl23Icp8E1m4TYFSQwCSF2GU3T8AcC+P3p99QyDIOvVpXjuKkfwX72+7XElKZvlLqrOZqPx9LsXQbw7A9r0YyW3VvM7/czc3U5lpP6vX11djG1btMGGsLhMF8vK+OWo4aQH94SbP26ygUTilhZXku1k3iPiKPw49rU24oUV8WoMpv3cXyfz4freRhG2/8cibZLhuSEEM1OVVUsI8SckggfLV5HTtBgytCuZBtu0rCUoiisr0w/VFUdt3FRaFvTp5M5Xl0PSTolNSZ2Cy/Lo6oq66vSz52KWA6OB02JER4q36+tYN28GJfs348Mv47jeqiqwhs/FfPDmgoO7N8p4fcvlmIByq3VxC3yAgpeE9cxUnWDiOLnzQWlLCldx6judWtc1a3DJE+ziR0jgUkI0ewso25+yqryLUMuz89ay4UTizisXxaKveVL3LZtRvfM4amZa1Jdin6dwmhN3fSrBfgUl5E9slldkXoC8tie2ei4tOTXtGVZjOyenba8V24Qo4mjhJ5rs3fvPP78/iJueXt+Uvk+ffLwq17C72CmX8OnqWmfrOsU9uPZTVsCQNUNllXDFa/MrO+9/GDRRsI+jftOGkGe5smwntghMiQnhGhWms/PUzPXJoSlze77bHnSEJDruvTNC9A9O/W8lcsO6IvhtP0tJhwzzmljemBoyQkk7NM4YnBXbCt9D9SuYNs2vXKD9E6xlQrAhfv1JUTT3tt4PM7YXjnkhpL7qVQFpo3vjV9JDEYBz+LUMT1SXu+QgZ0Jqk0PMqbq5/dvzEsa6q01HW5+eyGW1rLDo6L9kx4mIUSzMjF4Y+76tOUfLSnluEHZmOaW8KBbEe45fhj/+GQ5nywtxfWgMCvA/x3Yj96ZKk4DCw5qPj+mp6IrHopttup6PGHi3H/SSF6dU0xhVhDXg5KqKCeO6l43DJTmdaqq4ul+bA98iodrxZs8HLWZz6rhr8cO4x+fLOWTJZtwPI+umX4u3r8fgzsFicfTP2lmBMNEXRVDBd1OvahmMBTiue/Xc8uRQ3j8m1V8t6ocgD55Ic7bt4g35xRz5t69Er5sHCvOsXt2JuTXeOKb1VTHbQKGyvHDu3HiiAJooE66YWD+fDU/dtrFKDfWWtSaDkX5Ic7Zpw/ZAYP11THu/3Q5yzbVErEVMrbj/RNiMwlMQojmpZByk9TNaszk2OC6LoZZwxX79eDiiX2wXQ+/Bn43jp1m/SJVN6jBz1PfrGb2uio6Z/j51V496Z0daL21m1yHvFCQnjkh3p63AU1VOHpYIRk+FTeeJi4ZQVbVOjz2zQrWV8UYUpDJGWN7kqlZDa5evb0cx8Hwqrli/z5cOLHvz++tQqZqEY2mDiaaL0CN5+Phr1bx49pK8kI+Th3bkwGds1Ci267dpLCmIsprP61j6vBunDS6B57nsaEqzoOfL6ciajJt715J91AciwP7d2JQ1ywsx8XQVDqFDBQ39aClqqrYRpj3lm7ijbkbAJi6ZwH7981DM2sSAqaiKMRtlz9OGUp20OCJb1dRXBllj84Z3HXsML5ZVZb2IQMh0pHAJIRoVprnsFevXL79uadhW/v3y8dKMTTleR6YMXyAD8Ah7XwfTdNYH9e48PmZ9U+ArSqPMnN1BWeP68XRg/OavH/ZzrCMML958SfWbjWP6a4PFvNalwymTxmEuk3PiaL7eW9pBfd+urz+2KryKDMWbuSe44fRJ6Q3y1YbrutCrIaAolBYWEhxcTHRND1Yuq6zwdS48Nnv6oPvqvIos9ZWctKo7pw6qhBlq3ZYZpxJAzrz4eKNPPHNqqTrHT6kKyHVZet+IE3XWVjpcvWr37Jtbrn24D2Y0COIu81nxDLC/N8rc1lZtiUM//WjpbzyUzF/OXpIwnvreR5F+WHeX1zKdW/MrT++qjzKh4s3cufRw8gP69ipt6YTIiWZwySEaFaqHec3+/dNOZdndI9sCsN6k4ebbM3P7f9bnPJx+Ue+XkW8Ff4tqBsG7y8qTQhLmy0qqWF2cU3S1iRx1eC+z1Ykne+4Hn+asRhTbfl5NpYe4s//W5Syl/C5H9YScRJ/X23bZnBBJv06hZPOD/s0pu3dG3ebCdymGuBPMxYlhSWoC0FxJXHdLV3X+Xx5WUJY2mz5pghfr6pMWjIg7nj869OlSee7Htz5v0XEZL632EHSwySEaFau65Krx3n01NE89OVKvl5ZTlZA58SR3Th4QCfUeA1NHQyJOgpLS1N3D3jAvA01jO2stehTUHF03p5fkrb8tTnrGX1wHzb3m6mqysqyKE6a8Li2Ikqt7ZHZTPXTDR9xxUdxZRTPCKI58ZTvT9T2mL8h/crqM1dXcEifEPF43VCpYfh4fOZqbv/FUBZsqMZvaChArWkzuGsWT3+3kjPH9kDbanJ5temkXYIhbrtsrLHotlVmshSDN+etrJs8P6SA0T1z6uvy9tz1vDF3A/v06Ac/92OpqsraiiiW43HciG5M2bMQQ1OJmjZPz1zNR4tLqYk75OzQOyg6OglMQohm59o2mYrDFfv1wN6vN3gefizsWHWTwxJAYwtmp+jc2uUUGq6Xpm6z2rXnoSgNd/KrNL0hiqLg+jN4bvZ6XvhxHbWmw5CCTC47oB8FAQ22mSfV6HurJq6P5FHXa7O+Os63K8t5d0EJpuOyV69c8kJ+yqPJQ4qNrXieqrh7tp+L9uvLC7PWcsMb8wDYr38+dxy9J6/PKWbrt8rz6taBeuS0MfywpoKrX/mJTbUmPXOCnDehiKOGFtTVQaYxiR0gQ3JCiF3C8zw8M4Zm1qJZEewGnnTbUQHVY3DX1H0vqgKDuma2+Bo7fmyOHlaQtvzYYYUozpb3wPM8eucGUg5dQt1TZkG96d/ojhHixrcX8ti3q6n9ecL9vPXVXPDsLNZFFTQtcUnQLL+Wdu0mBRjTMyfhCUfbMjlxVA+mz1jI63PX16+t9O2qcq5+5SfOHNebgJIYmjJ9Gl0zUw83Bg2NzhmJZYZncdb4Iq5/Yy6fLCnF8Twcz+OjxaXc8OY8zhzXG91NfG975QZ5ZfY67vl4aX1v1uqKKDe+OY91lTE6hWXVb7FjJDAJIdodzY5x7SF7EDSS1/++7IB++NyWXe8I6haJ3K9vHnt0Tp7LM7pHNoO6hJJCnM81uWbyHknn+3WV6w8b2OT1pxRFYWPEYfa6bZ9sq+tc+etHS7G3WY/IdOGcffsQ9iW/t2fs3QvX8xJ6iFRVZeGGatamWK3ddFyembkaZ5ueMlVVuOzA/ujb9LopwKUH9Ev+YlJV/rdwA1Wx5N6qyqjFR4s3wja9dTHb5bWfipPOB3joixVErLa/GKpoW2RITgjRKEVRwAgQ8zRitkvIUAliYpstH0ygbp5Uvmby+OmjeWPuBr5fU0HXzACnjO5OJ7/XKk/IAWjxGu78xWB+WFfNG3M2oKoKxw0vZHCXEJpZmzQC5Nkm47uHePasvSmPWEQthwy/TucMA92M4NhN6yXTdZ1vl5WTHdA5fmQPhhRkYrsejuvx2pxivl5RhukpbB2ZIqbDPz5eyu1T9+TL5WXMLa4kL+zj0EFdmb+hmi+Xl3P0oBxqauqeSlN1nQ+XpF936+sV5Zj7JK7DVB6xePWnYv5y3HA+WLSRpaU1dM8Ocujgrrw3fwM9c/z02mprFNPT+HxZGUX5IU4a1aN+z7rSWpNnv1/Dp0s3cdSA3PpYpmkaa8qjaUfcauI2VTGb/J17W0UHJYFJCNEgRVGwfRn85aNlfL5sEx51wyanj+3BlMGdIN46z2Y7jo3PqeakoTkcNzQPHXCsGJ7VehNTPM9DjdcwrqvB2MIiFDwUx8JJM9FdVVVsLcA/P1rCZz8v2OnXVU4c1Z0TRnRDsVNvULsj9SnMDHDrlKE88tVK/vPVCgAy/Tqn79WL4YVZaNtMGNJVlaUba7nshR8Z1yePPbtlUx2zuf29hVTHbW48fFBCT5nieeQG0w9vZQV0lG1ab2gqX68oY+aqcib2y2dYt2xKa0x+99oc4rbLiSO74Xlbhtg0xWPv3rkMLczin58uq38SsWdOkIv278ecdZVoypbtVxzHIexveMjNr6mkXUlUiBQkMAkhGuQaIa59Yz6LSrascxO1HB76ciW6pjKlf1baxSVbwuZerra0lWrd2kl2o3OKHV8GN7w5jznFW4bM4rbLk9+uRkHhlBGdsaI7H0ht22ZEjxwufv5H1ldt6XWrjtvc99kyfnfoQEKay9b9hGGfyn798/locSlfrSjjqxVl9WW6qjCsWzbR6JYgZ5omU4d14+XZqYe/ThzVg6BiJ95D9+jfOcySjbV8tLg04fy8kEFuQGPrF6iOxWGDu3LWUzMTFpxcXRHlhjfm8sjpY8BOXHKgc4af7KBBZTR57tygrpl1Q46pt/0TIiWZwySESEtRFMpibkJY2trj36wi2qS97ju2iriTEJa29vysNdS6yfOIdkTd0gWRhLC0tUe+WknUTuxhskyLCyf2pUdOMOG4pijcetQQLMdNmMOk6zpVUYvT90pezXt0zxyK8sNYJLZDs2P84YhB5AQNdLVuGxyfrhI0NO48eiiGs02S0XRe+GFtytW5bdfjldnrUPXEz2HIM7lj6p749cSvubyQwY2HD8LvNLwafGNP8omOR3qYhBBpqarKytL0Xyy1pkPccSUy7QRd11ldmr6LI2a51JoOvrRnbN895q9PP6y3viqGDQlxJup43PrufP52/HBWbIrw7apyCrIC7FuUz5ziSr5bVcbkXsH6FcgVVePF2WsY2T2bR08fy6dLS4mYDvv1yydmOdz89jwe/uXwhC8b13XJMUweOW0MVXGb9VUxOoX95IYMQm6U+DbbyFieytz1qYMlwJx1VVgjuyZMLbfiUXpnhnnyV3vxzcoylm+KsGe3LIYVZpNBjFgs9fw7xfATV3ysrYzh11W6hA18bjxp8U3R8UhgEkKk5bouXdI8/g11QzQ+TQV54GiHOY5D54xA2nJVqZsrRhPm1TuOQ4/cYNryTL+etGaVoSlcf8Qg/vHxUuavr6ZPfogf11by9HeruPPoYXiug7vVU4gKHvv1zUPXNC57YRZ98sMYmsrbc9dzxNACzp9QhLbN4KSqqsT1MDe+NY+ftnqCryg/xPSj9ySsOwlbwuiKR7fsYNrFSrvnBNEVL2lKkhWtxQcc1i+L7NE9KS8vJxqtIO0jAb4QL87dyBPfrq5fhTzs0/jTlCH0zfThNsPefqL9kiE5IURanudRmOGrfyppW4cP7pq0xo7YPp7n0TnDl3Y9ogP36ExYa9oEdsdxGFqYnXL5BYATRnYn05f4NZBhqLw6ex0fLymlpCbONyvLWbChmo01Jpe/NJvskL9ub7qfxeNx9uyWw23vLKAyZvPj2kq+W1VORczimZmryfDr+LeZYeb6wtz5v0UJYQnqtjm59tW5xPVQwnHFsTh1bI+07Tx5dA9w0vcARSJ1vaSxWPqnJ3Vd56eSKI99szphy5Za0+GKV+YQUZrS1yd2BxKYhOigFEVJ2n8rFZ8T5e/HDaPLNosJ7t07h3PH98Q1W+cR/rZMURR8Pl+j72/QifDX44ZTkJXY0zS8eza/2b8fbiz13LHt5fP5WLmplluOGkJ2IHFA4cA9OtG/cwZVVmIXU7UNr6ZZv6gyarGstDZhTzy/38+789an3eLl8W9WEdtmMCPiwJdbTSbf2rJNtVTHt+my9AVZsamWCyYUJazdZGgKF+/XlyWltai+EE1hqj7+81Xy5sFQt7ff/xaV4vNJaOrIZEhOiA7I84Uoi3t8vqCMgFHLPn1yCKsOSor1ixzHIUeLcf+Je1IacSiPmHTPCZKhe82yL9zuxvMFKTfh8wVl6KrChL75ZGguipU8X8myLPICOv88cSQba+JsrI3TMydIdsDAZ9VgN3GTYl3X+al4E92y/Tx86hjWVESpilkU5Yfx6yr/99Js/nrc8IQ5aJbjEWtgUcdVFVGG5WfUD5l5KCxLsSnuZusqY7goCf86j5gNP89fFjHJC6v1PVmmq/DirHWcsXcvHj9jLMs3RVCUutXQF2yo5rnv17BPr+wm9QB4KKyrSj+nbGlpLergnCbcQbR3EpiE6GA8f5h7Pl3JB1s9zn3vJ3D2uF5MHZyf8ovdcRw0p5auOhTmqLhuLZiyFde2PH+Y+79cwzvzN9Qfu+/zFZw6psf/s3fWcXLU9/9/jq6du1uSS3JxhQgQXBJcSoEW2kKh1FvKjzpOaUsLhbbftlCKFIoFdwgWLO5udzl3Wx37/bHJ3s3tbgIkAZLM8/Hg8SC7szvzmdnbee1bXm8uGJ+LELGfW1GSqe23+MFTi1AlkZyUaLFxpkfhnvPG4xL6bXPbPi2RSITjRuSwozPIV+7/GJcs4ZJFekMaVTk+7jx3Qlx9kSoJSdvxAapzU9AGjbmxTIMJRem8u6U94fYjcnzIll2ApbpkJFFI2PUGkJfqstULuSWLK2dXsmBTG799aT1pu6JlvSGduWMK+NaMCtyStS/lXkiYDM9JSdq1OKk47XMft+Pw5cJJyTk4HEbIsszyxn6bWNrNvz+uoy1k7bWdenD9isMAkiSxrjVgE0u7eWRpPU39JqJo/8qNSG6ueSZq1tgX1tneESCim7T0hbntjc1YSvKi8E+Cruu4VZkbXl6PaUX9s7qDGqYFW9r8PLy4jtQhWcMUUefr0+MtAgBKMjwUp7ttnwFd15k9LCfhKBWAbxwZFTOD8cgCp9bkJ9x+SmkGXmXIrUkLIyDw8rroue0N6bExKS+ubUaRxH1ODYt6mO/Mrkj4nE+VmFGRaROKDocfjmBycDiM0ESFR5Y2JH3+yZWNyE6dxmfCkNQ9ntvHljcgKgPnVhAEmvsj9IUTF80v29lN0Ez+Fe3xeJA8qSjeVFtN0WBcLhcf7+hMWl/06vqWuBqmcDDACSOy+daMCo6syOT8ScWcWpPPjIos/nzOeFy6vVPN5/Px5sYWbpo7hsrsgTqiTK/C/zuhmiV1Xfgt+2eqL6RzRHkWZ4wr5Ohh2VwwqYTjq3M5vjqXi6aW0h2wx4oM2c0jS3cmPRePLqsHJXk3oKyq+MM6ipL8s22aJiU+kRtPi/pD7WZYjo+/XzAB91BvKIfDDicl5+BwGGFaAr2h5L+SuwIapuUY9n0WDIs9n9ughj4oOCcIAn0JhskORjMshsoAWZYJyz5e3trBu1va8aky504sojwjDSFkTyeJokinP3miKqybGJYV56PltiKcWlOALAosru0iN0XlWzMrUCXQhwg8UxDY2NrPC2ub+crkEorSPZhWtA7qqZUNbG/3c8a4ApvXU8SIDsbdnWbb0NxHWZaX08YUMH9FPfPG5JM9aJacbu753PYENTQLhn5yBdlFjynz1JJG6jprGVOYyuljCvBYocS+SlqIybkK9104AX/ERBIFfLKFbISddJzDwSOYnn76aRYtWkRDQwOqqlJdXc0ll1xCUVHRF31oDg4HDapgMK0skxfWJh6WenRVNqJlOCO2PgMqJkdUZFLXlTgScVRVFjJGrMHeNE3KM31J3y/do0RTU0P0Tkj2cfXjK2jpGxhH8+7WduaNKeDyI0sRQn2xx8PhMFPLM7n/49qE+xie62OIqwCKotCuKVz12BJb9OvVDa18Z3YVp4zKgeCAMDMiEWZVZfPOlnb+/NaWuH3MGZ6DW7R7JGX5VM6dWMx3HluOZkRF0bL6bp5b3ch1J46kON2DNUj8eUSTaWWZbGlL7MN0RHkmHsGw+SsJssLqjgi/emFlrEprWX03jy1r4K/nj6dQNROKIEPXkNFIh6i/WMQZOecQ5aBJya1bt46TTz6ZW265hV/96leYpsnNN9+8R18NBwcHO0YkzNemlcSNiwDI9qnMqHTqND4rhhbm/AnFeBN4HmV4FI4dkYM+5Nx6RZOTR+UlfL/vzq7Ebdln9CkeHw8tqrOJpd28sLaZ9oBhq5MyTZPCVDej81MT7uP7Rw/DI9pr0jTZyx8XbE6YKvy/hdsIDlEP4XCYKaUZcdYIEG37v3xmBdaQAc2GCXe+vSUmlmLHa8Gdb23BZGia0M85E4pJccX/xk9zy8wbW0goaO/U00QXN726Ma4xIWKY3PDKRjRp3+rDHA4/DpoI0y9/+Uvbv6+++mouv/xytm3bRk1NTdz2mqbZvvgFQcDj8cT+f0/sfv5wnCV0uK79cFq3zwpz31cncdc721hc14UkChxfncsVM8pxG0GML/AcyLJMWFCwEFAwQI/sU5dYMg7U9fZaIf711Ync8+52PtrRiSgIHDM8h6tmVSQ+t3qIq2eVMbEknZBmkuJS6ApGqMryMiJLwdBCtmMMWjKvbWhNuv9X1rdw9ZFF9PdH/ZsUReGt7e389rTRLNzajiqJuBWJzoDG2IJUtrb7KUlTUQbtI2BYLK/vTvj+FrB8ZzfHV/hsP1Y9RoB7zp/AmxtaSHUrSJJIZ3+EGZVZZCk6etjeTNCzaxxKIgKaQWtfmGLVfq58hPjnhZO4592tfLi9E0GAmZXZfPfoKlKsINqQc9vSH0lqj1DfHcSvW6QdBn/vh9N322AOxLoPGsE0lN3OrSkpKQmff/rpp3nyySdj/66srOT2228nNzf3E++joKBg3w7yIOZwXfvhsu584PYzx9IX1hEEgXS3jFeVgcSRiM+D9v4wCza18sjSLfSFNI4oz+LymZWUZXqQpQMTDD8Q1zsfuPWMsfSFNEAg3bPnc9sViJDhDXLfhzto6Q0zKj+FqWWZZKT58KqZtm3ruwLoSVrxATTTJDU1ldTUgX2Ft9WiGyYeReJ/S+vpDmpMKctgZmUWzX0hZEWhMKcwtv2Wtj2bZWqGSWZmZtzjDd0BDAseWFRHSDc5alg2XpdMRnoKimSPunU0JZ9vB9F6sMLCwrjHc4HrT62JRb9S3TKZXpVE53ZnbWJjzN2IokxhXvw+DlUOl++2oezPdQvWgfj5doCxLIvf//73+P1+brzxxoTbJIswtbW12WYUJUIQBAoKCmhubj4gv26/zByua3fW/cWu21S83LZgGx/XdtkeVyWR+y6aRKYQ2q92Bl+WdSO7eGJNBw8tie8A+8OZYxmTKdi+ryRPCne9V5s0yvS3CyZS5tZs50pzpfOntzbx/ja7gJBFgb9/ZRKlKRKR4CCR5E3nZ8+sYWNrYuH0yKXTSTH6bI/p7nSue24Nm4a8JsUl8++LpuDW7MXomprG5f9bRlcgPv2rSiKPXDoNNZJ82O4nIaKkcOEDSxIKzByfyj8vGI+sJa6JOpT40nzWP2c+zbplWf5EwZSDpoZpMPfddx91dXX88Ic/TLqNoih4vd7Yf7vTcRAVXHv775Nudyj+d7iu3Vn3F7f/1oARJ5YgWm/yl3e2YUrqIbduy7IIIfNwArEE8Ic3NxMShqxb1zl/UgmpCWp5ppVlku5WbNtLkkRXMBInlgB00+Lud7YQRra9RkDgqtlVKEOn8gKnjy1EFgXb9qqqsr65L04sAfSHdR5YVIviS7O9xieEufb46riuNoDvHFWFG22fz61qhvnuUVVx7y8APz+xGpcZ+sKv/+f135fhs/5lXvcn5aBLyf373/9m6dKl3HDDDWRnZ3/Rh+PgsFdkRSEkqGhm9Fe9Bw09El+0e7iiKAoLN3YkfX5xXRcaww6aLytZltFkDyEj+ivXLVrIejAusi2KIts7A0nd0lv7w/g1i7RBj0WQ+P0b67n1jLEs2NjKotpOfC6ZU0bnk+lVmb+igSum5hIORz9fPp+PD5Yl7ogEWNXYS1g3GTwlsC+s8/DiOv52wSTmr2xkZUM3WV6VC6eU0hUI88H2DuaOSIuVRbjcHt7YtDXpPhZubecbR5bbrAvCoRBjc338++Ip3P9RLZvb+inJ8HDZkeUUp8oYSWboiaKIIbmIWCII4BJMRD2cMPpo6RGOq0yjOnc8939cR0NPiJF5KXzjiDIyZB1d3xdfcIfDkYPlOwjLsvj3v//NokWLuP7668nLS9xZ4uDwZcJSPbxf7+cfH2ykwx/BrYicOa6Qr04qsrV/H+54krhEQ7TT6mApV5VUNy0RiT++uoF1zdHrO7kkg58cN4IsNYwWGbhJW5aFO4nh5G5kURjS027R2BPiR0+t5OjhOZwzsZiwbvLSuma2tPm5aGqJ7fWWZe2qn0pyvKLA0JpYSRQ4b2Ixjd1BMjwy500sxh/W2dDcy7yxhWxs7RsiUCzcCboud+NWpITXzxBkVjd2UpblYWJJOj1BjVUN3ZQkcQCXZJlOQ+WPr21hVWM0XTexOJ1rjhtOuhTGNBKUWmhByjwivzmhEsXtxQj5MbUgln74pKYc9h8HjWC67777WLhwIddeey0ej4fu7m4AvF6vM0Ha4UuJrKgsqO3jjgUD3jQhzeSxZQ3UdQW5bk4FRJIPLT1ciEQiHD0sm3+8vyPh86eOLkBFOyi8cPoslav+t4SIMSAoltV3c9Vjy7n/4imo2AVTSYYHlywS1uMjJCPzUvAqdsHkQueMsQU8uqyBtza18damNttrTh6ZR2TQvLre3l5mVWVzz7uJI0DHjsglRREZHPDMdMsENZMbX1kft/2y+m5unjeGULA79pi/v5+zxhfFxpYM5cxxhaTLFoOrhRSPj8dXN/Gfj+vitt/WHuB7R5VD0P6DIiB4+PZjS22dbysaevj2/1bwwCWTUY3EP0BM00TQQuTkZNLU1/mpUjAODoM5aGqYXnvtNQKBANdffz3f/va3Y/998MEHX/ShOTgkJCQo/DOJCPhweyd9+sESNznw+ESdK2dWxD1elO7msuklGAdBClPx+Pjv4jqbWNpNf1jnpXXNuDxe2+OGZfHT40YgDvkopLhkvnPUsDg/Ij0S5oJJxZRn2t8H4GvTSklXTJsgEEWRiG5w+YyKuO3zU12cO6GYsGZPTfVFDP5v4baEa1zX3EdbfxhpUNebokSTbaeNie9GGp7rY3JpJsEhfkt+U+LRpfUJ9/H6xlaCQ/42ZFVl/qrGhDYBAc3guTXNyIor7jkHh/3JQRNhevzxx7/oQ3A4BBEVF2FBoTdk4G/rx614UYxQ0jEIkiShSW76NAvdMEl3y7isCIYWXw8R0Kykc8IAdnQGmJAlOsNsAUELc8bobI4bmUeHP0JEN8lOcZHhkiDUy8FwhoKmyMqG5O3yy3Z2c9aYgVICURTZ2uZnQ0sfD1wyjVfXN1PXHWRScQbTyjO55dX1/OLEalsNE4Ac6efOc8axtrmPNza1kuqSOWt8EXkeATNk7/qSJImFG9sxLXjoa1PpC+uEdZMMr4JhmPzyhXXcfe5YWw1TSDdp7w/zi5NGMqYwjb6QjluRCGoGf1qwibVNvVRVp8dqmERJ4dEl2zmyIosLJ5fQG9LRDZNMr0pPUOP6l9Zxz/kTbDVMvSEd07K4aGopU8syo3VUssi6pl4eXbqT5r4Qw7wDfxsRS2LJzu6k53ZxXTfnjsk5aFK3DgcnB41gcnDY76hent3QwYOLdsaiAuVZXm6bN5p0ORxXpCvJMh26yi+eW0dDdzTt4ZJFrpxZwfHDMuLSa6ocrb1JlgDI9CpYluOqDaCqKm0RgV+9sDo2WsQli1wyrYwzxuTZRnF8WZGF6MiPnd2JR6NkeVUU0WL3Fbcsi4I0NxdMLuF3r23Eo0hkeFXe2tzKptY+fnXyaGTBgiGa21B9vLG5jdc2tDA8J4X+iMGtr23glyePpsDlQgsPROMsy6Iy20eKW+E3L61je0f0M6pI0Tqlr00vQxpSxKRIIv++eAovrG3m9jc2Yexqyx+Rm8L1p9XQ4bf/bQiYjC9KI8un8rNnVsdcyL2KxDdnVHDWhKK4VIZHkbh53hheWdfCT+eviv2NTCnN4JbTx5DuVjDNgR8hsmCR5VWBxDYA2T4FWbAOirStw8HLQZOSc3DYnyiKwoc7+7n3w1pbCqW2M8DVT6wiJMaPTQiJHq5+YmVMLEF0eOlf3t3G2rZQ3MR4j2Awqyor4f7TPQoFKapTT7ELv+Dhe0+ssM1hC+sm9324g4/remy2IF9WFCPEV6eUJH3+K5NLMEIDotqyLNK9Kr98fi0rGnr4cEcnL69rZlVjLy+vb+HJFQ2kuu1jcV0uF2tbAvz13W1sbvXz8roW3trUxtb2AN9/YgV+7J9bXdepzk/lt4PEEkSH+j66tB7NMEkbUgKa5ZJYuK2DJ1c0xMQSwOa2fn4yfxXF6R4ig4rX/X4/s4bl8OsX19lGtgQ0g3ve3cqI3BTSZLuUSVEl3tjQylub22w/KJbu7ObeD3aQ4RkyDliPcPEezu1FU0qwnK43hwOMI5gcDktCgsp9H+1I+Fx3UGNTW8BWpyHLMsvqe/BHEv+G/cf7O4iIQ+48WpifzBlGRZa93sSnSvz57LG4jMSRiMMNVVVZ1dhDbyhx+vLfH9XSbykJn/syoWkao/N8nDMhfiD45TMqKEyR49KvnQGNre2JoyYvrG2iP2LfPiC4uPfD7Qm390cMPtzeids9IJoURWFza3/MIHJ4jo/pZZmk7Oqce3jxTno1+22gO2zwv2WJ64ta+sLs7ArG6pYg2nizYFNbrHD94iklXDW7krRdM/Xu/6gWv2HfR1A3eXNTYvPNNU29+CP2z4JpmpSnK3xtWikC0dq2onQ3AvCNI8ooSZGc1LbDAcdJyTkclhiWQFt/8l+km1r7mZibEatlkiQp1iaeiNquAJZgvylYloWi+fnzWaNp7tfZ2NJPcbqbqmwPLjOEoTsJBIje1De0JB9j0dwbwrAgufHAl4hgL5dOK+XcicUsqe1ClASmlmbiky2sIV1fqqrS2JJ8eLhmWAQ03VZfZCFQ25m8s3Jjax/zRmbE5rwpisLGlg7OmVDEuROLqe0M0OGP8O3ZVUR0g2ufWYNu2W8EYcOkfw+1d1s7/IzLHRjSbIkS61v6+PdFk0l1K6xr7iUQMfj7RZNxySJXPLwUzRJsNUz9YZ09THihrT9Cus9e3ydEApw3Lp+5YwtZvatWbFxxenR4cBLfJgeH/YkjmBwOSyTBItun0uFPLJqG5fpshd+GYVCd50v6fiUZHgQr/heuaZqIYT/FqkBZhQfTNDEj/U6txSB0XWdEbvJzm5fiIoHp9JcSS/Xy9KpGHl3WwIyKLHTD5K4Fm/n+McOYU5EK2oBAikQiFKbHd7vtRhYFvIrEICcCBCxKM7zUdiUWTaPyUm31RZqmcdTwbPrDBt96xN6SP70sk3svmow85Ny6JBGfKiWNplZm+WLGmACiZfKDY4axor6H217faEvjzR1TwP1fnwpDoj8pLnmP9X05PtVWwwRgKR5e2tDOvz7YEXudAFw1u5KTh6cjaMnFp4PD/sBJyTkclrisCJdNL0v4XKpLZnSeXTDpus70skw8SuI4xxUzylHN5BEry7LQdd1JGyQgHA4zsSQDXxLzykuPKMfHl78+RRRFtndHeGDRTiK6yTtb2nl/eye6BX9+eyvtYSFucnq2V6U8K7FoOqUmH9+QTGSWS+DSIxJ/bt2KyBGVWQSDA6leTdNIcyv87JnVcS35i+q6eHZ1Exle+07S3RLnTSxOuI9sn0p5lsc2p9Pv9xPSTG55dYNNLAG8uLaZRbVdZLrttxqPLHDM8JyE+xiVn0rKkNEvoijS4Df45yCxBFHB9feF22kOWIiicztzOLA4nzCHQw5VVXG73XFF2IPRNY2jKtO5eGoJ0iATnMI0N389f3zC+iK3EeSe88aTmzJQq6RIApfPKGdCgXevQ50/DYIg4HK5cLvdB+xGIIoiLpcrtr8vEh8h7j5/IvmpAwkoWRS4eGopsyszYymmLzOm5OLBxdHan0yvwjHDczh6WE7s5v/Y8gakIV5BHjPAH88aR3VeSuwxAZgzIofLZ1Si++3dgR3haDrra9PLbLPeclNUbjythuU7u0hNTY09np6eztKd3Qm9oQCeW91IV9D+ue0O6pwwMo/TxxZQle3l2OpcppVlUpnl5c5zxtPaF7b9beXk5PDc6kYsoCrbw9+/MpF/XzyZU0ZFLRQeXbqTjpBdSAXCOvPGFjKjIotR+akcV53L+OJ0xham8Z3ZVfSHhnSPyioPL0lcVwXw36X1CLJjYOxwYHFScg6HDrJKEJVn1rXS2BNiRmUm4wpScRnBhL5KkhbkgolFnFpTSFt/GI8ike1T8KBhhOO3NwydfMXiH+ePoy9iETFMMj0yLkvD3I+O3ZbqpTVo8uKqFnTD4tTReRSlKoiR/TNZXRAEDMVLXW+EV5Y14ZJbmVuTR7ZbQIh8MYXoWjhMoWrxtwsm0BsyCOsm2T4Vj2BgBJN7G32ZMBDoCUa45vgRpLkVPtzeiSwK/OKkkTT2hFi4rZ2hEzkikQg+l8TvzhhLf1inP6yT6VHxqiJSuDdB6lbgvg93cOvpYzlxl2eVSxZJcyt8tKOD7R0BTh05MGNTlmWae5KLTX/EwDAtW32RZlr8++Navn/0MFY19PDB9g7GFKZy0qjhdAc11rf0UTksJfYDIWJatPSF+d83pkcNOtc2E4gYHD8qjytmVXLDy+sxsdeghXSTNza28OPjRvDxjk5WN/YwqzKbOSNyeHl9C1OK00j3CLEuUt0SaOtPbl7a2hdGtw6eEToOByeCdZj1Nbe1tdnCyYkQBIHCwkKampoOu7bvg3Xtgqyyql3j1y+ut4Xsc1NU/nreBNx6f9x6DFcqP5i/hvruoK2e4mfHDWd2qQ/0z99d2lK9/GtRAy+utY+ZmF6eyS+OH4YQ3vfiVsuVwk2vb2bpTrsQOWtcIZdNLUQ4DMa1HIjPuaS62N4PD35cx+K6Lttzx1Xncs74IqpSLLRBJqeiLLOlF655di2GaSECJlHbib+fP55UK2hP43oyaOgNcceCzWxt99s+txdPLWV2VTZVqcQicm63m3VdJj+evyrhMZdkePjLueORQgOfBcuTRsQQ+N4TK2gdJFIE4Fcnj2JCSTquyMA8uYyMDNrDFo8uqefJFQ22968pSOWW08ciCWD5B86J6E2jLWBErRAG1UopksAfzxpPeYYbYdAxyaqL+5e3x73/br46pYRLxmehRxKnbg/W77X9weG69k+zbkVRyM3N3et7Oik5h0OCiOjity9viCsibeuPcOc7W2FIKkRSXfz74zrqd3kqDX7dHxZsIcjn38YuiiJ1vVqcWAJYVNvFkvrePaYZPwmKovBhbU+cWAJ4ZnUTjf26UwvyGRFMg9a+cJxYAliwqQ3dtDCGDIiNiG5+8cL6WO3PbmnUE9S4+bVNGLLdV8klRPhge2fMimDw5/a/S3bic8m29GUoFKI8y0tZZmIfq+/MriTTY09lyZLEnW9vtoml3fu69bWNmBY2ERcMBmnvjyQUM+ua+3h5XTMZQ2qYNEvghpfXxxWWa4bFb15cGxdZswydM8cX4kow5Ncli8wbU4iVxJ3fwWF/4XwzOhz0SJLEupa+uILT3Xy4vZOQZRcaYRRe3ZDYBwbgox2dNq+ZzwNRUXliRWPS5x9f0Ygm7NsxhZF5PMmvdIAnVzYhKfuvFkRWXJiKB1SPzdfqYEIQBETVjal4EFV30novTVB4Ynn03E4oTufymRV8a0YFo/KjNUVPLG+w1dkIgkBDT5iglvhGv665j6Bh31dAF3h2VSOqBL89ZRQPfG0q/7l4CqfvmuP2+oaWWF0aQGpqKst3dvGHs8Zx9LCc2My63BSV35wyGsuCjoA9KrPbzylFlfj9mWN58GtTufeiycyoyMSwLFY19OD1DhSqp6Sl8dzqpqTn77nVTbT57UKxP2wktUfoCem0D+leNUSV+SsbuWXeGFu916j8VG49fQxPr2zAlJwaJocDi1PD5HDQIwgCgSQt0BD9ZTxUTFlEf80moy9sfO6F0KZF0lZugEDEwETYJz8iiz2fK3/E2KM/zidFkiRCkpcn1zTz/rZOUtwyF00uZlSuDyG8f2qxPg8ERaVLk3no/Z1s7fBTnuXl61NLyXGZWJo9AmMCqiTyx7PGsb6lj3c2tyOJAsdV5/L16WU8v7rRNkxXEISkYmk3+tCLIUTf75JpZTy7upGHl+zEq0qcO6GYrx1RzhPL6mybi6JIQ08IzbT49swKrpxdScQwUSURAYsbXt7AjXNrbPFUw7T4zuxKjhqey/+W7mRVYw9ZXpWvTi3lu0cPY3Fdty0KaRjWHj9TgYjO0IyInqQIfTchzUCQB2qYTGBtUy8fbu/kvInFfGtGBQJRZ/7fv7GJLK+KZeXv8T0dHPYVRzA5HPTous7YwqEjSgcoz/TilmBwnF/BZFxhGqubEs8om1GRiaZ9vjVMkmVwYnUuy5IMGT1meDYuQR86WuxT4RIMjh6ezRPLE0eyTqrORTD3rdtPEAT8oofLH1luGz68qqGHk0blcfWMkoOiTkqSFdZ2Glz33MCss+0dAd7e3M4Np45iSp6CoQ/UQ7ow+dGxw/nNi+tiqV6ADS19jMxL4ecnjUQyw7HrZ5omFdmpSf2Isn0qKaoIgz6GsmVw8bQyLn90GT3BgX2vbuxldlU2Pz52BFpk4DPt9/s5YWQeDy2u43evb7K9v1eR+MPZ4/AoItaggE6mR2ZaeRaXPbwk5t69vSPA0p3dXDi5hAsml9Df3x3b3tQjHDsil7c2tyU8jzMqsvGpIpFB60jzKKS45IQGmZIoUJTuwQoPHJSKwZzhOfzzgx3c8+7WuNecP7EYBR1nMqPDgcRJyTkcEqRIJieOjC/aE4CfHjcc1bSLH1EP8eNjh9ksBXYzrSyTHI/0uRdIaprGEeUZFKbFz7FLd8ucPa4waVHrJ0WPhLlgYjGprvjfSsUZHiaVpO2zPYKguPjH+7U2sbSb1za00hmyvnAbg09CWHRx6+ubEoqZ37+5mbA4pC5OFlla12UTS7vZ2NpPbWcgLi2pSALzxhYm3P+VsypRhnw+FUXlvg+228TSbhZu66C1P2w7t7quY1jw8rr4uriAZvDgx7Vxw3fDJvzhzU0xsTSY/y2rJzTk8f7+fsYWpVGZHe8n5VUkLjuyHGuIQPZIcPVRVQlWHS3gdg1xKtW1CCePyiXbF592y/GpHDcie6/NPA4O+4oTYXI4JBC0IN+bXc7E4nQeXlJPhz/C2KI0rp5dQa5qxokA0zTJVXTuv2gSf1+4g6U7u0n3yFw4uYTjR2QjhJKPQTmQKJqfv543jsdXNPLC2hYM0+L46hwunV6G2wjsF4dwjxHgvq9O4j+L6liwuQ1VEpk3poDzJhSiaH721VpTE2Te2dKe9Pm3trRz0dhM2wDXLyN9YSOhMIFo6rI7pJMz6CdnwJQTCpPdPLe6iclFw9kdMhJFkc3tfkbmp3BEeQ0WoMoi/WGdLK/K25vbqMn3kTroPQKGwNtb2plQnMaP54zAu0v4bm3v58ZXNvDq+hZ+fFQZ3d3dAKSlpfHiysSRH4g2E/g10zZ+JRgxWNfcR0Gam7PGF1GR5SUQMXhlfTOLa7tYWtfF6aOy6OmJNg7IisoL65q54+zxPLWigRfWNhPSDGZVZXPFzEpWNnQxrTgNeVCorN2vke6W+eNZ47j3wx1sbe+nON3DpUeUk5/qoqEnSNGQcj2XHuCfX5nAw0saeG1D9DyfUpPPRZOLUfXkfxuyLBMRXdERO4oH1YzsV8+03ecgLChYCEgYyEYkoZWJw8GNI5gcDh1C/cwpdTOjfAzs+uIS9TBmgl/KAKaukSHo/PzYMnSqcLtdEOpD+4LEEkSFnBTu45LxOZw/oQABUNExIn37bZyKYRi4zX6+c0Qhlx9REi0SDvvRw337LJaAaH5pD3MvxIMgugSwd1Of+A32tDRREBjsFGRZFoIg4pIl6nuCPLG8gQ5/hGE5Pr42rQy3IhHvLGTx4znDGFmQxt3vbGXZzm7cisTcMfn89+vTeG19c/x+95JHGLoHATi1poCjhmXz4KI6NrT0ke5ROGNcIWeOK6J3qKkk4I/oaIbFaTX5nDWhCAEwTQvNMqnrCjK92J4yFwTY0u6npiCNHx4zDJ9LJqgZdAc0tnX4GZ4gWiWKIrolUJ7l5ZrjqwHoCUYwEJEkKbFAcfl4p7aX/3wc7forz/Ly3dmVVGe5998oFXcqz65v47Fl9fSEdGoKUvnB0VUUeWSsL8CaxOHA4Qgmh0MKXdeRBlX57E0AWJYFkRCyIJCbk0ZTX/IhsJ8nhhaO/SI/EL9TLcvC2rXuvJx0mvrjW+E/Kyo6x1fn8lqSLsRjR2SjHQRzv9JcMllehc5AvEBIcclkeGRbfZFP0Jg3tpC73t6S8P3OHF+IZIRj19OyLKqyfbz6/o5YxARga7uf619ezzXHjyDdLWIMymalKjCqIJ3LH1kaKwgPagZPrmhk2c4e/nDWOPr6BgR/b28vRw3L4f8Wbk94TLOqsklRJVt9UapLYmZlFr94fm3ssZ6gxkOL6phensn/O6E6Fl2CaLrsrAnF/OipVXHpSJcsct9FU3ATsdUXZbgl3LLEz55ZHXdMZ48vYnZlJgTtYiMgevn2I0vpCdmjQ//5uI57L5qMMmR8jqC4eWhZk63ztLYzwLXPreW6E0ZwVIkHXdu3KKelern1jS0sqh34+1nX3Md3Hl/JXeeOY1iKvN+jWQ5fHE4Nk4ODw37F1MJcPqOcTG+8BcKZ4wrIUDgoDPREUeCHc4bH1fgIwA+OGYY05NvTFETGF6VRlR0/SHh8URoFaW6EITVMYcO0iaXB3PfhDiJDOjlDpsjf3tsa3z0HbOvws6W932YroKoqlgVnTyiK2z7NLXPR1FLCQ+Yb6qbFP95PLLAW1XbRP8QFX1EUVjX0JKzdCusm93+0A1O2fxY0Ex5cVJtwH8+ubozrYHV5fTy7ujFOLAF0BzVeWtuMy2OPSoWQeTKJTcc9720nJOybDYEgCHSGLZtY2o0F/OmtrUREx+rgUMKJMDk4fApEUcSU3QRNAc2w8CkiLiuM7hScxrAsC68R4N4LJ/LGpnbe3dpBmkvmwsnFlKcrSTvkRFHElFwELRHNsPAqIm4zjK5/Mee206+xYFMb9108maBmEtQMPIqEWxH539KdVGR6yB/0DRowJO5YsIHLZ1bQ2BPi3a3tSLtsAFLdMv9cuJ1fnjg8tr0oiuzo8JPmlrnmuBFU5aQQ0g1cssRH2zv463vb8EdMBiezNMNi6c5uxhamce7EYjyKhCQKbG7r58nl9by3tZ2jKkcQCETPcWpqKo99XE9+qpsbTqvh1fUt9IQ0JhSnc2RFFn98czO3nTGGwW0G/oiRUPzsZll9N6dW+QiHoxEgSVF5d2tyb6/FdV2EzEqbdUFfSE9qoWFa0NIXpsw9YCsQMEQ+2tFFeaaXCyaXxOY5tvSFeXxZPR9u7+SMMXmx9KIgCDT2hJJlhaNjaCIGGUmPeu9IksTKBAawu9nRGSBsCiS2DHU4GHEEk4PDJ0SSJPoFD7e+uolVjdHW7Wyfyo+OGcb4PDccBGmmzwvDMJCNPuYNT+Hk4RlIAqCHMSOJxY8kSQREL797YxPL6qM3oQyPwg+OqWJygWf/1Zt8ChRJ4ISReTyzqomX1jYTMUxkUeCk0fnMGZGHe4jr9O7Grl88v5bqvBSmlWViWhZPrmigtjPAUVXZKCKx1JRlWWR4VP711Sn84/1tXP/yekwLPIrEBZOKue+iKSgSDPWRuGBSCWWZHu55dysduwwexxamcf1pNWxs6bO5cJumSZZX5U9vbSbLqzBnRC5VOT42t/Xz6JKdKJIYF0GTRQFRIKkfV4ZHse1DxCLNndxQNcUlx9VJKUPDc0PwKBLWIK8DWbCYWpbBhOIM/vbeVuq6ooKuPMvL944Zxor6bmTBsqU7Peqeb2+KJO5TvtuyLNL3sG5JFJATdOE6HLw4KTkHh09IWPby3SdXxcQSQIc/wq9fWs+OXuOgdbI+kOiahqAFMSNDZqINISJ5+N6Tq2JiCaKplhtf2cjmLm2fR8J8FtLcMh9s7+CZVY1Edhkt6qbFS2ubeWVdM2ke+zG5iXDh5BIANrX2898lO3l0aX3M0fqCKSVYg4YbW5ZFfpqHW15dz4JNbTGBEtQMHlhUx7tb2kgbYv+Q65OYXJrB7W9sioklgDVNvfz2pXUcMyKXjo6O2ONdXV0cUZGFKEBnQGP+ykYeXlzHxzs6sYCTRufHpU7T3CJHD89JeE5kUWBsYZqthT8QCHDW+PiU327Om1hMhmK/9j7ZYlhOfOoSINOrkOm2/y3JRphTawr41QtrY2IJojVJv3phLaeNKYgT1TleiXR34s9NdV4KXmnf0sKGYTCuKDWhNQnA8dW5uPbJNc3hy4YjmBwcPgGSJLG2qc92kxrMXxduR5dcCZ9z2DOSJLG5PRA3u2w397y7fa+1IAdi/l1Yt3h1fbS+SCDqKbT73vjOlnYiuv2GayoeUtwKs6uyAUh1yaTsinKcNqYATTdxeVNsr+kJaTEB7lVgdP6AiHhseT3+IfvoCBjc9+GOhMfbFdBY09hDTs6A2HG73fQEI/xozggEIN2rxPZRnuXllNH5dA+xTvBHLM6bWEJxejRR51EkJCEadbr2hGr6w7rtx0Fqaio9wQhfnVIad0wTi9MZnptCv2kXLpIe4sZTR5HulpGB4Tk+3HK0SPz2M8agGvaUoCmIPLWiIWHtlmZYPLuqCUS7yFKNMLefMSZu/lyGR+H6U0Yh6XuOWn4SrzCXGeHG00YxVDOVZHi4cmZ5nBu8w8GNk5JzcPgEyLLMsvrkHXSbWvsxEfdpbMnhiiRJrGpMfm5ruwIYiIl/3XlS8WvQ1Bsiw+Mm0yPjNoJJTQxll5ug4KKlL8z27R0UpaXhFjTMUHxdVX9Yw6vKfPPIciqyfHQHNdI9Mg3dIe77aAfdQc1ew6Rb/OWtLfzuzDH8cM5wGntCiCIUpLkJhnXueGsLN542auBYZJmd7UG+MrmYCyaX0hPU6ApEKEhz41EkfvzkKvwRg8FS0bRgU2ty24ulO7s5rWZgRIjH4+GD1Y1MKE7hiW8dQadfoy+sUZzhwSNLXPW/xdx53mTbjSCkmdzy2gbuOHs8umHS0B0k3aOQ41N5eX0zfWGdM6rTY3VSpiDx9KomJhan8++Lp/Delnb8EZ2jh+cQ1k2uf2k9//zqJFsNk2maZLt0HrhkKn0Rnda+MFk+lXS3gs8KEwoNGcpriaxJ4soPsLqxB21ini31Zxo6JR6Bh782mcV13ezoDDK2MJUxBam49ABGkoin6PYRtCQau0O4FJE8nwsPYbRwvMCy9AjjslUevXQqC7d10tIXZnpZBlVZnqin2UHQ3ODwyXEEk4PDJ8A0Tcqzkpdv5qW4SGo85LBHTNOkNDP5uc3wKIgJzq3lzeCOBZt5b+tACqoo3c0fzhpHlirEGWOKLi+r2yPc9MrqWMGxLApcMbOSU0ZmwxD/La9L5uZ5Ndz7wQ7WNA2M46jOS+GmuTWkuhUYFKVQJYHbzxzLR7Wd/PXdbbE0nkeRuO7Eaq49YQQeWWR37MQwDGoKUhmRm8KPn1ppSzUdUZ7JneePx9BN7N3yFvmpbpp6E0dHSjM9tg5ETdM4dUwe/WGTK/+3PBYh3e219PcLp8ZFbVRZ5I6zxnHvB9tZsGnA9DI/1cXtZ45FN+xGsIoIMyqy8KoyP316FZXZPhRJ5NcvrmPumEIun1kRVyclSRJB0cMNL69nRcNAGnZYjo/fnTGWFNWwXT9ZsChMc7O1PfEcwqJ0j62GaTeGrqGiMafUQ+7kMlpbW9HDyT3NLE8aT6xq5r+Ld2LsOo+pLpmb5tUwPN2DEY4vhrf0CG4inFrlQxRTo+7qkf7942nm8KXCSck5OHwCNE1jdlVW0iLOS6eX4ra+3M7VX1Z0XWdKWRZqkkLgCyeX4BPttx/F4+OBj+tsYgmgsSfED59cSUCMF2C9hsQvnl9j687STYu/L9zGls4QimKv5XErMg8tqouLbGxq7ef/Fm7DrdjjiWmyRXsgwp/f2hITSxCtSbr+pfXoBujB/tjjUWEj8NOnV9nEEsDHtV38/b1teFT7PvJ9MhdPi099QbTIeM6IXFpbB/yv+vv7kUSRHzy5wpZOtoCX1jXz4tpmsj32dWe6ZZ5f02wTSxDtSPvRU6vI8Ko2MdPZ2cnEkgxufGU9XQGNZTu7+XhHJ10BjYcX15HuUchUh6QvXan88c3NNrEEUQ+q655bQ0iyWwQIhsZFUxOvG+DCKSVgJO+m3C3w9uS+7XK5WNXYz4OL6mJiCaAvrHPN06vpN5MXeEP0OyIcDjsO34cwjmBycPiEuM0wfzprLN4hN8ozxhYwuyLDmWX1GZFlmdoOP9efNhqfKjE818f4ojRSXTLHVedSmunFb9q/qvymxItr412tAdr9Eeq7Q7a6Jo83hadXNibt/Hrg41o0yT7DLxAxWJpkEPK65r64tvg+XeCBjxN7C1nA48vqUXyptsfb/REaehJHi97c2Eb/kH20+jVSXQqn1hTYHvcoEr8+ZRTb2vx4PANiMSMjg9WNvUlb+B9fVk/nkBqmrpDOM6sS+xd1BzW2twds+/D5fLy6PjrGJxEPfFxLV9gueAOayQfbOxJuv7XdT+9QvyXVQ11XgCtmVtqKrCVR4KpZlezo8COq+9bAH0DlgSTeULpp8dqGFrzeeAdyh8MHJyXn4PAJMXWNyhSFhy6ZRENvGH/EoDLLi0fQIdy/9zdwSIgoiiyv70QQRf5x4WSW7eyiK6jx7VlVCAL8v2fW8J+LJ9rqw0K6ScQwkQSBWcOyqczy0RPSWLCpld6QTkNPkOo0dywSoiNQ3x0kN0XlvIklVOb4sCyLTn+EJ5Y30NATQrMEW53NYJEhiwKpbpn+sB4zVewLa6QO+gaNGBZNScQPwM6eIIHIgKhIS0tjZV3yz41hWQQ1w+aRZCFw++sbuP3McXx1Sglb2v2kqBIlGR4WbGqjqTfEkSUDkRhVVantSOx7BdAT0tFNy3YjiBhRz6lk1HUHOKI0j2AwGhWzBJHtHX6yfSpfn17G6PxUDAu6AxH+/VEtjT0hDEuw/Tr3R/Q9JrA7AxGyUwbGnURMgSdXNFCZ7eMPZ42jrT+MAOSkuHhxTRP13UGml6bvUwTAIurdlIwdnQEkKW8f9uBwsOMIJgeHT4FpaMiGRoVHQPAKmLojlPYVw4gOat3eGeTrDy2ORYHu/6iWmoJUbjljTFwNk0eWmFaWwdeml7O6sYewbpLpUbjuxJGsa+6lMtuHPmiOl4zBzIosLphcwr/e387fF24DoDjDw1WzKtne4UcVLdteUt0yGW6Zy2dVkp/qpr0/TLZPpSMQ4d4PdpDhUW3eW25ZZFiuj51JTB9H5qaQ4hLp36Vfent7KclI3FoP0Y4xr2ofvyJicevpY7n/41pW1HeT6VWJ6Cb9YZ0fHjuck4rybOmyYDDIqILUBO8eJS/FFZdmdkki6W45oas2wIjclFjBNwCmweyqLC47opw/LtjMn9+KjoYpSnfzk2NHsKPDjyLYr1+KS0YSBFvqa+hxGcZATZlbshiW7ePVDa28vqGVTK+CZRHr8Dt9bAFuyWJfkuKSYFGZ7WVdc+Ki+jGFac6Yk8McJyXn4PAZsCxrj75CDp8cwzDITnHxxzc3xaXM1jX3sWhHJ94hNUxpksFVs4cR0U18LpmWvjCaaeJVJUbkplCYqtquTzAQYEZVNr9+YS3rW/oozfBQme2luTfEb19ax+xhOUiaXeikKBJ/PGc8z61u4mfPrOb2NzZx7bNreGxpPbedMZYU1f716UbjsiMq4lrMIWqCec7EYkJ99pqdTI/CqPzEgubs8UVkDPERynDLbGjpY9nObkwr6gPWF45Ga+58awu5Ke6YAzdAX18fI3JTYs7YQ/nGjHJyvPZ9ZHkVLj2iPOH2xRkeSjI8tn0Eg0EmlWbyw6dWsnZQvVdjT4ifPbOaiaUZpA4p//GpIqcM6uYbzOSSDHwue9pbD/q5ZFoZNfmp/PbU0fz8xJH84qSR/PbU0YzOT+HCyaVEAvv240U1Qlw5qyrhcz5V4uhhObGomsPhiRNhcnBw+EJRFIW3tnUkTdE8vbKRc8fl21NyhkVEN7jt9Y22YuZHluzkuhNHYlqWbXtFUXhtcxsnjsrjjHFFbG7rJ6Kb1BSksmxnNw98VMtPjy5hsK12UDe5552tbGq134h3dAb445ubuGVeja3lv9+U6AuF+N0Z47j9jYHjKkxz86uTR9HYHSCzIBW9Lzp7TFVVOiM6V8+uYm1zL+VZXgRBIKwZdPojZPlUesO6LU3YHbF4fHk9FVlezp1YTG6KC1GI2gm8sKaJNza28M3pJTHzyoyMDJY1+bn7vInc+Mr6WPTEq0h8fXoZ4wvTaQsM2UdIJzfFxdeml/H4snpMy0I3LMYVp3PFjEpCukm6PDBUNi0tjWfWdcTXHRGt3frHwu385pSRtseN/m6umFmJIMDL66L1TwIwe1g2Pzl2BEqk32b5aFkWGW6Bb86osF3zHJ/KL04aRYZqoe+jltE0jfL0FH518ijufmdLLMJWkeWN1tcRwqlSPLxxBJODg8MXiiAItCUxBAUIaAYm2ASQhsR9H22LMxI1LfjDm5u4/+IpZEgDNTCWKJPlVXHJIpc/stQWyTpjXCGnjytAR7LtI2KYcV1cu9nU2k9IN4d4JEVTTG9tauVPZ49HM02EXet7cFEtX5lcahMBqqpS1xREEkXa/WH+83EtYd0ky6vwtenltPaHCUQM0ge9xrIsppZmctTwHO79YAe1XQFEAWZUZnPbGWP5aEeHzRVdlmVWNvTy8OKd3DyvhqBmEtFNUt0yDV0BLn5wMY99Y7pNMIV1k01tfRxfncfJo/LpCETwqRKKJBKKGKxq6ObEqjT6+/t3nVuJ5fXdiS8esKGlj7BhMbTHTAh08Z2ZFXx9Whl9YR2fKuNVRQj0oA9J1QmCQE/Y5NpnV9uuXbs/ws+eWc2Dl0wmhX1HCPczu8TDxIum0BvWkEWRFFXEY4aIhJ0u2MMdRzA5ODgcUERRxLIsmz/QYHRd54iyDP67ZGfC50fnpyIPcbXRTFhSFz8lHqLOz9s7AkzJHRBMiixRmePjm/9dGrf9c6ubmFKaiSrLGIPuiaE9FD5DtIsubZDCUmWRZ1Y28vaWdhp6gkwrz8K0LN7f1sGm1n6yfS6+M7uC3U5ChmFQlunl7ne28tGOAePOzoDGXW9v4fKZFaSqMgyqQ1YlgVPG5PPT+atjETnTgve3dbClrZ87zhlvO8+hUIgjyzM5Y1whr65v4akVDfSENCYWZ/Cdo6r43elj4uwcXJLACdX5PLy4jjc2tsb2U5zu5hcnj2JMYZptH7IgUJRu7zAcTF6qK+n4EN3fjVsQ8Moiph7E0hJ/RkRF5b8f1SfscjQsi8eWN3LltDz0yL47a4dDQRSCZAOYQIh9qo1yOHRwapgcHBwOCILLS7+cwrIOi60BBd2ViijFe9mYpklphivhbDEB+OExVciG/UZoWnuuH4sYpn1ArCjx9MqGpNv/d0kdumW/qXtVOWE90m5Sh8x5i+gm7f1h7jx3PJceUc64wrSoMJldxe/PHMvOrgBhbeCYJEnCsrCJpcH8b2k9xpBEpW5aPPBxXcL0ZUtfmO3tfttID9M0yU9zc+urG/jH+9tp90fQDIvFdV1c8cgyMn0qKS77bUCRRV5e38zrg8QSQENPiF+/sBZVEm01TJGIxjHDc+MG7O7m7PFFscHEgxFFEVP10Wa4WNpm0BhRMVRfwpmMmiWyqS15jdLG1j40y7mdORxYnAiTg4PD/seTxt0Ld/D6hgETRY8icfuZY6lKUTE0+292KeLnj2fW8ODiel5c20LEMBmRm8JP5gyj0GNh6PZoj0cWKUp3J20DH5WfiqEPdDvplkVbf/I4Qac/gsmQuidJ5MRR+bF5coM5elhONDKjD95e4Jrjq+mL6Dy8eCdL6rpi6bJLppVx1axKFFmIvSQcDlPblVz49Yf1uJScYcKaxuQjQj7e0cmJI3MHjklRaOsMs2SXn1R5phefS6KuK0h/WOeut7dwy+ljbL+cg5rJc6uaEr5/Z0CjtjPAhJyBGiYLk40tfdw0t4aIYZLqVmIjQRq7Q6iSiLmrRmk3giAQUXxc+9w6m3t3QZqbO88ei08M2ASvIliUZXqSOn2XZ3qRBTOpg7eDw/7AEUwODg77FdXl4oWN7TaxBFHH65/MX8Ujl07DNSTJYVkWYqiPb03O45IpxZgWKIKJbITjxBKAKIhcMbOSG15eH/fc8dW5RHQzGsHYddNVBJhenslHOzopSnczqyobWRRYurObTa39TChORxbsw20E4LSaAlRJ5OV1zeimhSQKnDgyj7ljCmDouA8BBFHg2mfWxHyMdqfL1jT18rcLJsKgKJYoiqS7k0dFBIiKskGaShIh06vQnqTmqyjDbUuXKYrCRzuambHLUmFbh5+eoMbFU8voDWnc/c5WIrpl83rSTZPAHtKRDT1BphZmEQpFxaquG4wrTqc7oPGX1zfGhih7FImrZleS7lGQsGxJVUt2c9sbm+MEUHNviJ89u5a/nF2DGBl4ztTCfG1qKW9tbo+9NxA7zxdNLcbU9jxM18FhX3EEk4ODw34lKLh4ZOm6hM/ppsWHOzo5qcIbN+sNwNDCyIOMh5Ldtvs1g6V1Xdx17gTu+3A765r7yElxcdGUUrJ9Kpvb+skrVGKCKazpHDUsh3SPgmZYLNjUimZYnDQqn2/OqKAk3YOC3ccnbBiYlkVxuptbzxiLYVrIosDWXamhsG7vLjMtmL+iIaHpY09QY8HGNi6cXBR7zLIsCtI8ZHoVugLx/VfTK7Ki41cGPaVKIl+dUsrd724lx6cyMj+VsG6ysqEb3bA4Zniu7T0EQaAqx0dxhoefPbPaNjeupiCVm+bVxKUdXZJImltO2PUGMDw3xV6PZpl4FYXvPL+csG4fCfPnt7Zw17kTME37OQlaEkvquhO+/87uIH0accXuOS6Tv5w7Ds2E/nD0/VJdEookkKmYSeufHBz2F07S18HBYb9iIiQUALvZ2RVAVeN9gURRxFK99Is+uiwPupqCpCT2D1IkkWOrc3l5bTPTyrP47amjuWRaKYvrOklxyYzKT7XVwqiKREcgzBsbW/nd6xtZVNvF8vpu7nl3K39/dxuaYRIZUsMUjBj86oW1iKKAIgpohokkCqiyyC9fWBu7ae8mpJt77BZbsrOLkDFwU7csi76Qxh1nj4+rhyrP8vLT40YQSRBdm1yWwR1nj+e7Rw8jy6syLMfH7WeM467zJqCIgq2GKRwOU52Xyq2vbYwbsruuuY/3t3XgG+InJYsil0wtS7iGwjQ3WV41bgzQgs1tNrE0mH9+sB1JtO8jqO9Z3HSHdNs6IOrE3dQb5ufPreW3L63jty+t4+fPr6Wld98LvfeF3Z/bgBT93GpqCpLi+kKPyeHA4ESYHBwc9isi0anzyepNxhWlx91wJUmiT/Dw2xc3xHyPfKrEVbMqOaoiFSFiN9nxKiLvbe3glQ3x9UXrmvv4v69MwjQHbqQmIp1+jQ+3xxdY13YFeG9rB2eNL7Sl5FJcCiHN5G/vRV3BVUmMDdUVgHSPAoOOyyWL0ce6EhsCZXgU3JLI4EElXlXmzrc288uTR9Eb0mjtD1Oe6SWsmzyypI5Lp9sNJEO6SW9Q55/vb2PjIH+ox5bVc8WMCvw5Psoy7TPVltd3J53z9tK6Zi6eVma7EQQ0g3SPwk+OHY5Llkj3RGuSNrf2c2RFFuua+phdPCBkJUVhQ0tid2yA2s4AYVOw7cOrRiNDmpH4uLJ9KtYgp3ZBEGgJwW2vb7JtF9ZNbn5tI/++aBI5opC0E/NAIYoiIdnHTa9uZNWu2jK3InLp9DJOrc5CiCQfS+Nw8OFEmBwcDlMEQUBR9jyB/bPgJcLVRyV2TM5LcTG2MM3WZQUQkb1894mVNpNIf8Tgjre2sKYtbPMWgmhH2ku7hu/m+FQml2TEuuw6/BFa+kK2MRaSJCcs3t7Nq+ub40SFADE36qJ0N2ML0yjJiIqROdW5kMAr6KKppSTjK5NLMAZ19ymKQk9I58MdXVz33Br+8f523trUxu1vbOLmVzfw/OpmIkMEhSwKLKrttIml3fzrwx1k+ewROUVRaOtLHoEJaWasQDv2GlGgOxghxSXzn49rue65Ndzw8nq6ghGCusGIPHs3o2BBZVbyES9lmZ64tJ8iCZw1rijh9rMqs1El0RZhEhQXDy9ObDsB8N8l9YhfQFRHk7384KnVMbEE0XP6j/d3sLC294D8fTl8cTgRJgeHLyGyLKMLChYWKkZcRGZfsVQvnWGL9zd04lb8zKjIwCcaCPuhcNYwDAzT4toTqrnvwx0xc8kppRlcdkQ5DGmKlySJNa1+OpOk8f6+cDt/PacGYZAACmpR88UfzhlORDfZ3NZPjk9lWG4KjyzZSWNPiKqiQakjgThhYDtma+hRRQfEHjM8h0umlbGtvZ+Nrf0My/FRnZdKW2+IvrCBZ9BPTt2wKMnwcNGUUkozPWT5VASgsTdEf0gn3aNgDSn6bukdiEB0+CM2I87o8F3dVpBtmBYvrEncwQawcGsHE4rSYv8WBIExhWlJt6/M9iIOSX2luCQK0zz85qWBOrSwbvLMqia2tPm5cW4Nlj5wnKZpcPzIXB5dujMWgRvMRVNLccmCzSVbN2BkfioXTi7hmdWNhDQTRRI4cVQ+Rw/LQTNM1EHXS7cE6vcwGLe+O4huCUmtDQ4EoihS2xumqTfxcd33US0zysYjO/7ghwyOYHJw+BIhCAKG6mNxUx/zV9ahmxZza/KZUZGJovn3y/w6y+XjL+/VsmBXxxHAPe/Ct44s5/RRWQjavs2YCAsqt7y2ktIMD987ehgeRUISBdY09vKrF9byzSPLObHCGxMziqKwpqk96fs1dAfRkWwF1qoscMNpNdz2+kYaBg27VSSB35wymuIMN6I4qP7HspgzIpf3tnYk3Mcxw3JQJcFWZJ7qllEkkaseW05JhofCNDcrGnrY1u7nnvMnkuaSbKaSggCvrm9hWnkmdyzYTP2u4xqW4+Mnx43gjQ2tXDilJLa9KIoUpUcjVqoEP5gzgqJd+3hwUV3C4bsW0B9J3sHWE4y/OXuU6Hy9zQl8jC47oiJOZPRHDP763taE77+mqZfWvjCl7gExI0gSpmVx49wa/vDmppjoc8kilx5RTmmGl5But2wIRDRueXUDx1bn8ptTRiMKAqIA725p51cvrOXP54wnxT2QYnNLAiPzU5KmeUcVpKJKsJ9/V+wRSZLY0pb4eAC6Ahqa5dxkDyWca+ng8CXCUHz89pVNthD/uuY+yjO93Hn2GMRw8lqRT4Isyyxq8tvE0m7u+6iWIysyyZfja0EEQcBUPAQNIRrdcUm40RO2cltEb9w9QY01TfGeQY29IURxYJCFIAiUZnqTHnO2T41r4XfLEv9butMmliDq8n3Laxt46GvTsKxBXV6WRWGamzGFabYBsRBNEx43Mndohg1JFLj/ox3ccNpo1jT2sqMzwOSSDL55ZDkPLqrle0cPs6/bsjiyIpsfPrXSlt7b2u7nJ/NX8ZfzJtjOazAYJNPr489nj6M0y0dXIII/onN8dR5njy9iXVM3XkVisGenJApMK8tManZ5REWWbR+BQABZFLj6qCpe29DKmxtbiRgm5ZleLjuynLouP+OL7BGoiG7Ssoc03vqWPiqqfLGUpwB8tL2TKWUZ/PrkUWimhWFa+FQJlyzyzOpGvj61xPYeHiUamluwqY0Fm9ri9pHulmzriBgW88YU8ur61rjUqSQKnDwqP66o/UBjGAalQ+rFBuNTJRSn6OWQwhFMDg5fEiRJYm1bwCaWdlPbFWDBlg5OqfKh78PPaE1UeGTJjqTPP7myiR/OKEAbVGMkiiJh2cebm9uZUpqJKIo09Wtsb+/nmKpMhLA9ciERNZ0EOGdiEVneaGpq94DYKSXptvoiXdeZWpaJWxEJafERtIunluKTBVvLf1A3eX9bBxOL0zlzfFHMl2fh1nZe3dDCjs4AYzMHiywLtyzyy5NGsmRnNy+vbcawTGZW5nBKTT5tfeG4sRsR3eSUmgJ+/txam1WAa6nIDafVENYMBss8VRR4bnVjwgLrsG7y7pZ2vjG9NBYwEkURryJQkO7h5lfWx+bWSYLAyTX5fOvICryKwGCJbFkWX59extKdXXEF0yNyfeSlumxCQ9M0fKqbfyzcRk6qi+tPG40gCLT3h3l4cS0/Oa4aWRJsHkkuWUQWhaQCpCDVZbt+kXCYIyuz+f4TKzh5dD7TyjIRhajQn7+ygStnVeEW7ZYNHsFgVlUWC7fFC79hOT5SFWzzSCzgrc1tXH/qaO55d2tM0BWmufne0cN4a3Mbl0zIS3i8BwrTNKnKSiHdLccG9Q7mgknFuC2dxOYMDgcjjmBycPiyICk8s7oWiN60xhamIYkCa5t68UcMnl/TzHFVIxH3oSbCEiR6Q8lf3xWIgCgzOA9kyB7ChsWxI3LpCxv0hjRyfC6KqnJo8Ycp9rgwtIHtZSPML06qprEnxN/e20ZDTwhRgNlV2dx57gRyvRJGaEAGWJbFzk4/d507geU7uziiIhtFEmnoCbKppY+ReVGvocHyJ6KbfGtmBR5F4u53ttAZ0FAkgRNG5vG7M8bSGdAQsweqfzTDxLDgv0t2cuLIPK6aXYVhWnhUkUeX1HHuxBJ007R9IVoW/GnBZvJSXZw5rpCCNDet/WGeX93EH97cxD3nT7Sdu7BhsL45eQRwdVMvYc0uFCOixa9fXGtL7RiWxUtrm3HLIt84wt4l55JFMjwKfzxrPI8s3cni2k68qsypNfkcX51HlleJS9uuaerhFyeP4vk1Tdz+xiZ6gxoTitO57sRRdAcihDXDlu5UJIHjqnN5bYjx6O79V+X4MLWBdYqiiCoJ3Di3hrBuYgKablKdl8JVsysZnuuLr23SQvz02GH4IwbL6wcGHA/L8XH76TXIut8m4hR0ClJdPLColm/NqCDDo4Ag0BWI8O+PdnDmuAIU9Li/DFmWCcte+sIm27a2k5uagk+OOsvvj/S2ywhy93njueaZtTHDToBTRudx1tg89FDycS4OBx+OYHJw+BIhAN84spyagjQW13ZhWBZnTyimqSfIi2ubE84Q+zRIosCkkgxeSdIxtjs6EDseQcCSZDBNfvL06lhdDsAxw3P44ZzhWIIFmj2F448Y/PKFgaJh04J3t3awpd3P384bb9tWURTSPZDuUQnpFj+ev4q+kM7k0gyuPqoKw7KwBHtBb4pbxiNL3P3OQK2NZli8vK6F+u4gvzp5FIYxcKxuWSKohZlcmsEvX1iLf1cdkCwKXHpEOQ3dQSYWpxMZlGEMGyan1uRTlunlwUV17Oj0U5rh5ZLpZfQEI4SGRJgUSSIv1cXO7sQ1YIVpbhRZskUcOvyRpHUwL6xp5iuTS2xF3xHd4r4Pd7Ckrot5Yws5Y2whEcPkrU1tXP34cu44ezxFaRm295lQnMlP5q9i9rBs7jxnPIok0tIX4vbXN/L16eW4ZJFBDgz4wwan1hSwvSNgq3tyySK/OWU0G1v6mVHsIRCIFqxbCCze2UVZhpebX91Afzi6QkkU+NaRFSiiSE2ex1bDZFkWUrif3544DL8u0O6PkOlVSFWIiqUhYkbXNI4dns2TKxq59bWNtueK093MrsxCG5KulmUZv+Tl58+utdU+TS3N4Jcnj0IK9eyzDYFhGGRJYf7v/LF0h0z6wjr5qS48ggGOWDrkcASTg8OXBUPjytmVPL2yiZ89szr28FMrGphZmc11J47EjbZPPTeGoXPW+CLeSmA0mO1TmVCcjmkM3NJFUURH4CfzV8XVtbyzpZ0sr8oVMyvsYy9UL395ZT0lGR7OmVBEUboHi2i67PUNrWxqD1CTLmEYA2muDI/K9S+vt9U8La7r4vJHl/GPCychCYJNLJqmyUOL6xKucXVjLwHNwG0NnKmwGb3h3/rqRozB3VdmVIDcPK8mLoqliAIpLpnrB41fqe0KcMurG7hiRkV0bMmg+mvdtDhvYjFLd81tG8q8MQUMPuWyLNPckbzzK2KYBDVjyNgSi3e3tqMZFv9dEt9m/9bmNmZUZNoea+oNsr0zwPbOAA8Nac2/98Pt3HnOOJuYEUWB619az9VHV+FTZba09ZPtUylKd/Pw4p2cPrbAJmgs02BCUTrfemSZLR1pmBb//GA7vz11NKoY79puWRZCJEAKkOoVsCwNIrZJMDYUzc9fzxvH82tbeGldC4IgMLcmn3k1eciaP+51YdnL/3tmDds77F5IS3Z2c+fbW7hmTiVWcN9qAiEqmiTDT7YIOV4BS3e64g5VHMHk4PAlwbIsgprJ8wnaxj/Y3sFJo/Ioyouf5P5pUE2N/jDcfd4E/vXBDhbXdSGJ0RTMVyaXIAsC+qAwi6Io1PWGkhYBv7i2ma9OLWWwA07YEinN8HBcdR73f7yDLW1+FEng+Oo8/nDWOJbu7GJKQR79/dFf4KIo0twTTlggbpgWd7+zlRvnjrY9HtYMuhN0hO1ma1s/uflyTJQposBbm9psYmkwjy+rZ+zcGvu5kkX++cH2hNv/Z1EtJ9fk21SAacHoglS+NaOC+z/aEauJkkWBH80ZTmmGB2FQ+E7TNArT3WR4FC6ZVkZVto+wYeKWRRbVdjF/ZUNcl5woRA00NSNxp9zuWq7BrN5VE1dTkMqJo/LxqRKb2/p5aW0zdV1BDAubYPLIApNL07nttY2kuGSK0930h3UaekIoksDPTxxJKDSQRhMEgXc2tyc1x3x06U4mF1QnfG43nyTSY5omUriP82oyOH10DgAudPRwX0KR1R0y4sTSbt7b2s7VR1XZxOj+4PM2znT4fHEEk4PDlwRRVnniw/qkz/9vWT0TTx2OkKCMVFJdhFAI6dEbrgcNPRIvcizLojI7hZUNPRxbncNVs6MGk0t3duIP65RneWFQRkmSJOq7k6cWIoZJSDNsgkkRRU4alc91z62JRYU0w+KV9S2sb+njN6eMQhw0KkNRFN7fltxUcmVDDxHdYrAloyJJSIKQVABl+1y2LjndMNjRmdx1uaEnFNde3xvSCWsmx1fncsKo/Njzb21u440NrXT6I+QPUhrpbpmVjb0094b40znjaeoJIYoCuSkunl/dxKiCVCozvbFaZlEUyVIVbjt9DHe/u5V7dtU/CcBx1bncde4EPLJgS5e5ZJHTxhSwurGXcycWk+qSY3VuT69q5JTR+babtiiK5KSoXH/aaFp6w8xf0UB3UGNSSQa3nj6Wx5ftZKh7kRTxc9XsKra2+6nrCsZMMiUxatnglix7algQ2dqRvL2+sSeEgcC+Sf0B9EgEaddZTFZQLYoiHXsYmWJaEIgY+10wORzaOILJweFLggFJB54C9IV1zEQ3Hncq/13eyPxVTYR1E5cscv7EIs4bXxDXwYbipq0/6uLc7o9w1ztbUESB40fm0eGPkJviIkNR0XaJrUgksseWf68iRQfEDro3CQI8sKg2Yb1VbWeAnpCO7hu8fTT1lQyXLMY5RauyyDEjchK2pKe6ZIrTPcgEYwN+FVlmRG5K0nb8YTk+LAubdJAEgd+cOpp1zb3c8PI6Qlr03J5aU8CNc2uQJXvPeF9E56FFdSyr7+bFtc1k+1QM04pFwlRJ4KfHjYhtL4oiAc3g929uskVCLODNTW2kuRUumlpiE4qaaXLOhCJG5afyz/e3xyJ/k0sy+NPZ41AkwZYuM02TySUZ3Pn2Fj4YNBbm3a3tfLC9g7vOnYBLsqfBdF0nRY3wx7PGsaG1n9WNPeT6XBxZmUWmW8IM9GDDMhlfmMa7WxJ7aQ3L8SElSbTJsowmubEEEcEyUc3wfjFpNU2T/LTkn1tZFPCpkq0Tz8FhbzguEQ4OXxJky+CY4dlJn59VmYU6pBJEUD38++OdIAjcdvpYfnfGWG47fSyaCQ8taURQ7L+hDVHGAv767jb+9t42VjX0sHRnN79/YxMvrm2Odpu5Bl5jWRb5qS4qs70Uprm57IhyfjRnOGeNL8KrSFw4pYQU1S52dMNi3R66xT7e0WkbvhuJRJgzIjfp9qfW5OMeYmjTH9Y5Z0IRI/NSbI/7VImb541hW0e/bSyFbsFRw3NwyYm/8i6YVMIQ/UOqS2JpXRdPLG+I2R1EHa8bWbCplXS3feyFaUHboE6pDn/EljZs7gvbWvV1XSekmUnTRi+sbcIaGveyBHZ0BLjplQ22NOmy+m6ufXYN/ohhizAZhkFfWLeJpdj+TYt/vL8toQO6Hgnh0fuZXODhG9NLmTs6h2wxFC+WiF6/Y0bkJkwHAlwxsxLZjBdBgsvHig6T789fy9n3LeKKx1fxdn0Y3KkJ3+fTkqqKcR5TuzltTAFead+75BwOLxzB5ODwJUHTNI6qyiJnyDwwiAqB8ycW2dr3AUKmxFHDc2npCzF/ZQOLajt5akUDHf4wR1ZlEx4SRBYQ+HB7B7Vd8TfppTu7afeHGWzvI+xyYL719LFcNbuKZTu7+N+yenqCGneeO4ETR+XFt2cL7DFilO1TbbPhQqEQPkXih8cMi9u2IsvL16eXIw2JV8mSgIDA+ZNKuPX0MVw5q5JfnjSKX58ympBmkOaWbWk/lyQQ0Q1umltDcfqAIMzwKFx34kiCmoEk2r8OdQteXpc4VbhgUxv6kHWLgsDoguQ3+zEFabYxJKqq0rwHg0jNsAhpdoFsWlbSYvcOf4QdHQGbYJIkiUW1iaNqAKsae9ESjDOBaJTGCPnR+rvRA70276XBKIrCxpZebppbYxv8m+6W+elxI9jc1ocu2sWl6nLz9vYefvnCulhXYVt/hN+/sYn/LmtEdCWPDn1S5Eg/N5xWw8zKrNhjkihwxthCvnlEGUbQ6WJz+HQ4KTkHh88BSZKQJAnTNJPeeABUPcD/fWUiDyzaSWNPEAGBLJ/CN44sx2MG4zqNDCFav3LmuGLe3drOzq4A5VleZlflsKqhm9IMe+TGsKyEaazdvLGhldnl6bEMmyzL9Gomz6xs5LHlA/VVzb0h3t3azi3zxjAyL8X2y0vZdVN6ZGl8F5cAzKjMtt3UBUEgpJvMrMpmalkmr25ooSugcdSwbIbnprChpZfxBfZIkiqJ3P9xLUvquvAqEvlpLnpDOh3+CKok8vCl05CkAXES0i38EYP3t3ZwyfSyXTVOFmHd5L0tbZw5vgi/ZtjSX30hLWmNlAV0BzUKBgVVBAvOmVi8S0zZX+dRJI4faTdWNE2TgrTkA2MVScAtS3EGjhtakkfv1jT1cOrILNtj3l2RHwEYlZ+KzyWzrb2fzoCGNDTX+RkQRInXNzaxtd3PV6eUkpfqwjAtdNNi/ooGNrf1c0zlBFsq2Y/KP95PXFD/5IoGzptYzL6O0jVNEzncy3XHDyOgDyegGfhUCZ9kYgTjI2UODnvDEUwODgcQURTRFC8b2wOsauyiNMPN5NJ03GYYM0H7sSa5WV7fw6k1+Szd2YVuWEwty2R9Sx+TClMRdfvNUhEEhuX4WLCplVnDcqI+SqLAGxtbmLHLAHKwD4FlWXvs5DEtiOgDskySJMJhk8eXxxejG6bF39/bxh/OGmu7uWmmxeTSDNY297CyYaDzTRTg5yeOZF1zLxXpA6lHVVUJBw06AxprGnvwKRLp2QorG3pwKxK5PheGYI/+hHWTJXVdAAQ0ezdUxDDZ1u4nr2SgUMoti2R6FI4ZkUNrf5jHl9UTMUxmV2Uzb2wRmV4VdUiEKVmKaTdeVbZ1ybnkqMC5ed4Y/vH+ttgxjcxL4cpZVfgUCUUSYmLUsizS3QqV2d6EabnTagpwKyLWIMEkCpCb4kratViS4Y1z+j6iIotjR/Ry5vhC1jT20h3UmDumAMuC5Q1duDHihPhgRFHc4+dGwKI4w807W9r5w5ub4p6vyvYhDokQ9oaNmBfWUEwLmnpDVHriR/R8WkzThGAfXkFgWFEhTU1NGBGnk83hs+EIJgeHA4QgCIRkH997wu5hpEoid54zjlKvjDko2iRJEt2awLqWPm4ZZM73wKI6jh6eQ0Wml3xFto8VsaIF0BOK08nxqbGRFhOL05FEAc2w/5G7JYFjR+Tx8JLEaZ2TR+ejDtJYmqaxvT16x55Vlc1Jo/JRZZH2/jDPrGpka7sfzbTiogGyCOdMKOHCKSJrm3pJc8tU56UQ1Ewsy7TdCBVFQQyb/OfjWlY22H/5P7G8gT+dMx4BbLfcvc0N6w1rhAeNdwmHgwQ1k58+vZriDA+zKrORpajVwN/e28bVR1XxlUmFdA/SLbIoUFOQmrAeqyrbhyoJNsHUHzFJdcnsNEzOm1hCTkp0JExTbwjTspBlgeAQkRDUDW4/YxzXv7wuth8BOGFUHl+fXp6gA1Hgq1NKufPtLXHHJAkCs4Zlx6VIVUnkqGHZ/GT+qpjVwZMrGhiW4+N3Z4wFK7H4slQvfRps7wiQ5VUpSlVRjaDNPyt6bsPMG1PAo0vqExb6X3pEGYoZsXW0KXuJbLkVCfbJcczBYf/jCKZ9RFJUQoJKb8hAlgRSFQHVDO8x7eJw8CJJEprkpl+LdiyluyRcVgRDi2+3sRQ3t7+5OS4aEDFMfvbsGh68eBIyA3UUouKiuT3E0ysb497r3S3tHFmRRX5lOgyKMglC1EF6yc5ufvbMajr8EXJ8Kl8/opzqPPfQmbWEdYMjK7N4c1MrTb1208SxhWmkuCTb+A7DMHDLIjfPG8Oapl7+8OYm+sM6pRkevja9jO0dAaQhO0lRJT7aEa118qkSFdk+QprB397bBsAjl06Pda9BNILR1heME0sQFUb3f1QbtSIY9LhHkShMc8etYTdjCtKIRAZa3UXZxTOrdgDQ0B2Mi5g9t7qJE6rzbGkjQYCfHjeCG15eT13XgNdCUbqbn5800laPBNE6s1fWNzOrKpuQZtIZiCAIApNLM3FLIst3dnN8dTZ9uw7LNE0UUeSaZ1Zx7QkjSXXJBDSDVJdMbZef7z+5gr+cN8G2j7BhkeqSOWlUnm10iUsWue7EkSza0clZowdScoqi0B02uPnVDXGz8ra2+3ng41qumFEG2M+j5U7ltje28HFtV+yxdI/Cn88eS56iYQz6fhMEAUGAa46v5s63N9tm3M0bU0BRupuh4senSgzP9SV0Oc/yKmR7VaxwYsd0B4cvCkcw7QuKh3fq+rjnvYHhnAVpbm6ZO5p8FduXisPBjyTLdJkuPtjcRVWOD8O02NoRQBYEJhd6IWJPq4RMkSV13Qnfyx8xaOgNUzEo7SCpKi+tTRz5AXh+dRNHVmTabuouReLpVY38b+mAAGj3R/jTgs18fVoZ504qstXASKLIfxfX8bMTqlnT2MM7W9pRJZGTRueTm+Liva3tjD2iKBYNMAyDEXlp3LFgs63Tamd3kFtf28i1J1Tjc0kMboLqC+sx801/xGDtEEPK97a285XxA7VVpmmyuC55YfLy+m4My7IJJp8icM3xI/jp06vjtj9pVB4pqoxiKLEok2lFJ94nQzNMTKwh4zsE/t+za6LpNJdEc2+I/FQ3Yd3kVy+s5a4hYkY3TaaXZ3PN06uZN7aQKWWZWJbFW5taeXtzG7eePhbNGBBZoihS2xmgrivI955YkfC4gkPmvFmmyT/e38ZpYwr58znjqe0M4FUlMr1q1HyzKM0WAZIkiRU7uuPE0m5e29DKpUeU2Wq3JNXFA0sbbGIJoCeo8YOnVvPARRNtQl9VVe5f0kBrf4TfnTGWtv4IId2gPNPLotoubnhpPX85u4bBQ2HcZphfnjSKH81fRc+gTkK3InLD3BrcaHtMEzo4fBE4gukzIkkSm3s0/rDAHhpv7g3x3SdX8tAlU1BJXpzpcPChKz56esIsq+/m/xZuwyIabfjWjAqaAhalbnu6TBuUGSnL9FCS4aU7GImlXnpDBoJ3QDAZJvTtwYepN6wzNOcR0UyeWN6QcPtHl+1k3thCW0pHkkROrsknpBlUZfvIT3MjAGluhd6Qxsmj8kEYkA2yLNMRStyWDnDfhzuYVpZpu+EKgpC0PgWIRV52Ew6HSfcoSbd3yWJcFCuoW+zsCnLP+RP4x/vbWdfcR45P5eKppZRmegjpBp5BPgFuReD46lze39aRcB9HD8/Bp0i2WXL9EZ12f4RbXtuAR5HI9Cp0BzQCu34c9QY1PIMUlmZYrGro5sa5Y3h4SR33frgDSYh6XN04bwxL6ro4bqRdKLrkPddJDV23T9Q4d2IJf1+4DUkQyEt1Edaj9V8C8ONjhxMIDHzvmKZJf0SPvdfYojR8qszW9n5a+sJEDDOuTiiEwnNrmgGYWJLO1NJMOvxhnl3VRH9YZ3tnkJFpYiz1ZwkCXUGN97d18P62DrJ9Kook0toXwrSivlhD7RG0SIg8t5e/XzCR5fU9bGnrpyTDwxEVmWSoAsZ+GFni4LC/cQTTZ0SXXPzf+/EFjgAhzWThtg5OrvTtFxO2z4Isy2iCggm4MNG15O3LDntHlmU6wgY3vbLelmJr7Alx0ysb+N0ZYylJ84I+EE3xKBKjC1L5xhHldPgjNPeGSHGl8+1ZlTy6tD469X1Qes0lwlHDsllW353wGGZUZOFWBLRBN/WuYCTpSArNiJomDnajDusWw3NS+PWLa+MKjY8sz+RbMysxh9ykt+3BxbnDHyGkm3bBBIwrSouN5BjKlNKMuDqbo4bl8H8LE3dNnVZTgCqLthqYkGZy59tbyEtx8dUpJVwxs5LuoMbjy+pZ09TLjafVMLNkoMU9rFuML06nOi+FTa32dvIsr8J5E4sJ64bttu4e5NkU1AyCPXYR6FElWw2TIolEDIsHF9Vy0dQyvj6tLHqsusnd72xl3pgCJEGwJaeK0t14FCkWoR5MTUEqLkW02VmHg0FOGZ3HyoZuPtjeGUtJSoLAr04ZhU8ybNtrmsbU0gzqRudzak0+S+q66QlqHDuiEpci8szKRlxDTCU1w6I808tvTh3Ntg4/S+q6yE918eDXp/H+9nYae0OMzvDErqGp68wZns3bm6PGlR1+e3p6enkmqmDEWVfqoQBpksTRFanMGZaJaRqIWgg96ETmHb6cOILpM2Ii7nHUwpqmPk4bnvaFCCbLlcLHjX08tmwHIc3k6OHZnDW2AJcRiCvYdPhkeL1ePtzWm7Q76YGPa7nl9DFxLty/OHEkG1v7yfCq9IV1clJcBCIGF08pwbIsRHHgl3o4EuGIiiyyl+yMu+l4FYkzxhUiDBEae4tQuGTRFpVyKyLPr21O2JX1UW0Xc8cWUpWhxrJ4hmGQuYfojwC4hgyhFQWBr00v57pnV8elgobn+uI8miRJoqkjyA+OGcZf3tlqe64sMzrAN6QZti+r3e3+rf1h7hryGoCQbti8nmQRNrb0cdPcMby+oYXXNrQQMSxmV2Vz9oQitrX7GVuYZrupu2SR8cXprEpQW1Wdl4JHtgsmAYvRBam8u7Wdbz2yFEUSsKxoHdZx1bm7ankGiEZ2LG6cW8N1z62xCd90t8y1J4xEsBIURwe6ufb44XQFDZbVd5PqkplYko4bDStsv66WZVGQojC2MI0fPbUq9vjza5ooz/Tyh7PGIhl+W/rLLcON82r48VOrbDVi9324gxtOq2FErhdDHxDRuq4zvjCVonQ3jT32WihVEvnWkWVYWuLvSsMwwBgYnOtYSTp8mXEE02dEwKQ005OwaBFgdH5qvKHf54Cp+vjdm1tt9QcPLQ7w/Jpm/nXhRFTDCXV/FiRJYlVjcu+W9S19DO2AFgC/ZvD48npbVCPdo3D9qaOj0+4H/ZgWZZlHPqrjt6eO5plVjbyzpR3TtDiyIouvTi3lPx/t4AfHVNmiIKkuieJ0Nw098cXP5ZleUt2SbTZc2LB4Y1Cx8FBeW9/CEWXptnUXZ3iSRkFmVmajSPabuoXF+qZefnfGWB5cVMeapl7cisgpows4ZngOxpAWdVEUkSWRnqDGvRdN5o2NrXT5Ixw9PJc0j8y9H+7gx8cOt+3DLUePq6E7cWHwmEK7w7NmCjy/ppnWvlquOX4Ex1Xn7hIzJne9tZmtHQH+9dXJNsHbF9K5fEYFd729ha3tA3/n5Zlevnf0MHpCGjmDlq4ZFkFNZ96YAs4YV8iH2zuRBJhZlU0gEh0WXJxmF01BLWqP8Kezx7O4rouW3hDVeSlU5fj4cHt7XCH6boRgL1nA3GEpWJZFJNybsEMNotf8Tws2xz1e2xXgocV1XDktHwb9kFIliT8t2BpXUG9acP3L6/nfZdPj0niKFuDuc8fxn0U7eWV9C5phMbUsgx8cXUWqEEnoJu7gcLDhCKbPiMvSuGJGJf/vuTVxz6mSyNHDc4hEEqckDhSiKFLfp8UVa0LUZO+hxfV8e1oeRoKhrA57JhKJUJrhSfp8tk+Nm3eGKPDI4p1xKaCeoMZNr6znnvMn4hp8IxFEWvrCXPvsak4ZXcBNc2sQBYFVjT388vm1ZHlVrCGjUn2qzC2nj+UHT66wzaFL9yjcPK8GrywPHvOGZZHUjBGizw0erSqKIgLw21NH85sX1xEZ5ApdnO7ma9PLMLEXZGNZTCzJ4PdvbGLu2AIunlaGbpi8vbmNhxfV8u3ZVbZaL1mWkYQIqxp7eXRpPVNKM3CrEn9buI3WvhC3nzkubs6bVxW49vgR/OipVXFCYd7YAryKhDaoc1GVog0ZH+3o5OrHV8Ste1R+Kooo2CIcqizy6/nr+PbMSnJTVNr6I+SkqHQFNG54eT13nTvBFmGSJYEPt3cyoTiD1r4QuSkqFtG0aapLYWubnzEFA0LOsixMCx5bVs8Ty+uZWJxBhlfh9Y2tbHqvH5cscnx1ftJrpSgKYTM6Z0+SpITRY1mWWVjXnVRMvbK+hcumlTD4U+LXSVrrZZgWqxt7mFGo2jodoyaRfVw5LZ/LppVgAYplIBohTN2JGzkcGjiC6TOiCzJ94QDfPLKChxfXxW4k2T6Va46vpqEnSHWq/LnaC6iqykvLmpI+/+amVi6bVoSEI5g+LX6/n9m76mwSCY7zJxWTIpsMTjyEdZOF2xIPJO0MaLT7I6R5xdiNzrLg2OpcFtd18cyqRp5ZZbcXOHtCEZIo2G5+nYEIHhnuvWgKa5t62dreT3VuCqMK0jBNg+5QhMEyzyMLnDgyjwcWJe7GO21MIfKQ5W1o6eeZlY387syxbG/30+YPMyI3BVUS+eULa/nLeRMY7MOdqli09Ib4fydW8+SKBv67uI4Ul8zcMYWcOCqfTI9KcJDTsmmauBSR8ycVM7k0gxfXNtMX0phcmsG1x1ezdGcXVdn2URlh3WJDcy9//8ok7v1gV9F3isrF08rIcCtEDBPNGkiHRwyLs8cXxZ3T3Xx9WhniEMHkViSK0tz84c1NKJJAmluhL6QTMUwqsrx4FHsqUjctTh6dz5X/W05Vto9p5dEuuaeWN9LWH+bu8yfaxKggCDTvigyaFnG1a2HdJKwbcR5XkiQRlr28uqmdt7e041VlvjKpiBFZPoSIPeItCAI9e2gk0AwrLm2qm/GPDaYnpCMIiX24jUjYJr4cqeRwKOEIps+IjsgfF2xmZmU2t54+Bs20kEWB/rDOvz/cQUWWl5rZRZ+7H5MyxK14MEN9Yxw+Oaqqsqihm5+fNJLfv7HJFmk5elgOxeke+sKmLX2S6GY0mO5gBMszqM3c1KnI8iYsTC5MczOzKhtMexTBsmBNSz8iUJXjY3JJBl3BCOuaeonoJuOGDB/VIhqnjy3klfUtcfVYNfmp1BSkEgoPpGIMw0ASBZbVd7OsvpvKbC/pboU3NrTSvqvOaujnqicCY4vSue/D7eSluPjJcSMIagbvbG5jZH4KXYEw5T5XrOU/FAqR4/PwyOKddAU1LplWikeR2NDSxy2vrucPZ4/HIwk2p6CgZvL393eQ6W3g/IklfGVyCd0hnSd2pT+vP200RxYMlKJLQnQO2w/nDOeed7baRO9Z4wvJSVUxLfvtvSeo8YuTR3Hds6tp6AnF6sryU13cOLeGrqBG7qClq6JAhz/CHWeP518fbOfBRXUIRIuef3ZCNWFNRxAGjsmyLLK8yQeASIKAKos2WwiAkOTlysdW2urcltR1ceLIPL47swRhkL2FrutMK8vgnx8k3seo/FQUwb5utyxSmuGJzXgbysTidHQ9sf+Vg8OhjCOYPiuWRaZHZcGmNhZsaotzIj6iPGOfbf0/LeFwmNNq8nlmdeIo09yafFRLd/xNPgOCILCx1c/O7iC3nTGG1r4w/WGdYbkprG/u46ZXNvC/r0+2vcanCPhUKWmLfUWWF9MciAiEAv2kulP45pEVbO/w88bGVgzT4ujhOUwqyQALJC1o6xZLkU3mr2gg1a1wSk0+LjlMSDN4eV0zmm5yZGkK1qAXyGYYXfZy17kTeG1DC29uakOVROaNLWBGRXZUWOgDN2JBEKjK9iIJAoZlxRWLj8xLid7UB/U2mAh887+L+f2ZY0lxK6xr7iXdrfCjY0fwwpomNMPkO0cUxQSTYRjU9wT59qxKVjX08NyaJvpCOlPKMvnD2ePZ3t5PpifDtt/dNTFdAY1/fhDfXafpFpI0IF8ty+Kx5fVEdJM/nj2O+u4gEcOkKtvHotoufv/6Jn5/1jib4PW5ZH7w5AquOa4aVRap7w5SlObGAq57dg1/PneCTcyooklphpf/fBy1WrhoaikAqxt7eWtTK+dPKkExtFj8RZIkvIKUsFgaotFGYYgYlVQX9y2uj2sKAHh9YytfmVREnjxgVWFZFrkeicml6Szbaa/BEwX48ZxhyEbY9p0gi/CjY4cn9LiaUZmFWxaHJIYdHA4PHMH0GXGj8dUpxfzprWiHzlBpdOrofMLh5O3YBwLLssjzipwyOp9X1tunrBekublgUhFG2Cn6/ixomsZxI7K5+olVfLyjk9wUFbcs0dQbQjct5gzPQUa3fQ7cZphvzaiI6/wCmFqaSZpsxUUPMmSLbbrJotpOThiZhygILK/vIs2tcOKILPSQ/Ve/EAlw09zRXPf8Oq5/aX3s8dH5qdwybzQMud6GIHLvB7WIInx9enlsH4ZpcceCTcyqyub4MretYaGtP8QVsyr5v4XbbO/lUSS+PasSTbd3sIlYZPlUfvDUKryqyNiCdLqDGpvaolGzX5w40tY96vP5+GB1A69vbOXbMyv42QnVWFbUTPJf729j4bZOnvjWETbrArciUpbpsTlwD2ZMYSrWIKVoWhY9AY0Pd3SyYFMbBWluZFGgqSeEYVnk+NS490hRRWZX5XDNM6vxqRLZvmj9Ul9Y59SafHyKiDHo+vl7e1A8GZwzoZjwrpotSRSYMzwXVRKQJZH+/gHRous6HpfEdSdGo5b1gyI6U8syOX1sAUPjxREUXtvYQjJe3dDGtybn2MbCiFqA355UzfPrWnlieQO9IZ3xxel8/+hK8lQTQ7cLelPX6A1p3HP+RP723lbWNfeRuct6Yc6I3OgAadNp/Xc4/DjoBNOrr77Kc889R3d3NyUlJVx22WWMHj36cz8OTdM4ujKTJXXZvLt1oEBSAK47sZoUUf9CEvhCJMBVR5Zwyug8nljeiD+ic8LIXGZWZKJofqem4DNimiaFKQpTSzNYsrObtv6BO6VHkbhyVgVodoGsaxGOq0pHEofx7w9r6QnpKJLAqaML+OYRJYjh/jihbYb9TC3wMKZgFI29IcxdLeluNMxQvNi1LAsl0s/v546kV7PoCGjkeFVSFQsx3BcX5dQsiSV1XbT2h3lhlzmh7f2A4yrK2d2+p+s6oqDS4Q/z+zPH8tqGVtr6w4zKT2VWVTb/fH8b159q//vzEeGbR1Zw86sbCERMFtUNNCGkuGQml2YQDg8IB0EQyE9z0+GPcNvr8d5m6R4lLp4hIXDN8dX86KmVcWnPcyYUYVnYRAO6xuxh2Xy4I2rA2TykA2xaeSZu0bJ5JBn+Hr42rYxUt8xjS+up6wriVkQumlrKVyaVQCi+qcNlRcj0KQTCBsePzEMk6tfkU2VULcBQmaGKJqsberhkWhlZPpXeoEaWT2VTSx+BiEGKqMVVHO4pcG0lKO+2LAtCfZwzMp25o3KRVRUzEkLUExdky0aY4nQ397y7jatmVZKd4iKsm7yxoYX/LdnJt6YVOsVJDoclB5Vg+uCDD/jPf/7D5ZdfzsiRI3njjTe49dZb+fOf/0xOTs7nfjxCuJ+fHF3ON48sZ+nObnyqxOSSdNyWhqV/cYXVQsTPMJ/Iz+eUYiIgmRp6uM/5jttHxIifX500nIXbunl8RQP9YZ2ZlVl8bVopXjOUsHVaiAQ4odzH7IqJUYNHScQjaBihvqSdS3o4iEKQCreAIAiY4Z49XjvTNBEiftKBDK+AZWkQiY96QjT6k+FVaO1P/PnM8qqIghVL0ViWRWGqyhsbW3lpbTPHjMilpiCN2k4/P3qqnpNH5ceNsQiFQkwrSeOKmRU8uKiO8K6bckmGh5vn1eCzgrbAWn9/P9PKMlEkwTaHbDfnjC/CLYm2faiCSV2Hn39cOJkHPq5lbXMvOT4Xl0wrQ5EEXIqIGRo4a6FggCMqssjxqbHaq924ZJGvTy/HitijVaZpIoV7uWBiEafVFBDSTdyyiFcRIdSHnqArTQsF8Koqbo8XnyohCAKKaCGE+xPWM4phP6eNyWdZfQ8PfFRLdyhCSbqHS6aVUZbhIhy0izKV6A+gF9cmjjKdPCqPSJIuWE2LIAkaBTlpNDV1JW31NwyDQpfEtceP4N4Pa1nZ2EOWV+VrU0uZWORD+Jwj5w4OXxYE6/MutNkHfvGLX1BZWckVV1wRe+zHP/4x06ZN46KLLrJtq2maLewvCAIej4e2tra9FmILgkBBQQHNzc2fuA5pt0newT5097Os/VDg06xblmXCKFgIuAQD4yByUZdlmSUtGr95eUPC5++9cCL5SsR2DiRJosN08aOnVtEZGPibmlCcxg2njEJMkuaV3V78pkx3UEOVRFJdIh4zlNDM1fSks7a5j1te2RAbPQIwZ0QOV8ysJFuK2NrYAQx3Or9/YxMjcn2MyEulKxDhjY2t/OS4EeQrWtx+BLePPk3g3x/t4K1N7RiWxeTSDH5wzDCy3BIkiODtRhTFWOv+/vZXEwQBQfUStER0w8IlC6iWjhlJnG6MKClc8dgKugL29R07PIcfHlWGqCUfWvtpPueCIIDsQkNExEK1tIP2++1w/V6Dw3ftn/Y7PTc3d4/bwEEkmHRd55JLLuEnP/kJ06dPjz1+//33s2PHDm644Qbb9o8//jhPPvlk7N+VlZXcfvvtn9vxOjh8Wenwh/nzgi28usEepbhqdhXnTywmLYGzt2VZtPaHaegO0tYfoSLLS26qiyxvfO3PZ6W2009Dd5AOv0Z/WKcsy4siCQzP8ZHlS9xN1twb7V7b1t5PToqLkgwPxeluxCTdou39YVp6Q+ixrz2LonQPuSnuhNt/WWnsCfLimibe2tyOV5W4eGop44vTyU5ynhwcHPadg0YwdXZ2ctVVV3HTTTcxcuTI2OPz58/nnXfe4a677rJt/3lHmA4VDte1H27rNhUPvRp8tKMLjyozrTQdj6Aj7CWVLAi70oQHwMVeFEUs1Ytfj3oBeRUJtxn+RHMQZVnGNM1PdFySJGFJKl6fj3B/zxc273FfkRWViLXLuNLUPtHYo8Ptc76bw3XdcPiu/UBEmA6qGiYgrs022WOKoqAoiWdgfdIPjTVkhMPhxOG69sNl3UIkQDpw+ohUcnJyaGpqiq59L687kOfGMAwI9g0YbRo2t4I98mlEj67rCIZBanYG/V3aQXu9tUg4ZmfyaRNlh8vnfCiH67rh8F37/lx3cpfDLxlpaWmIokh3d7ft8Z6eHtLT0xO/yMHBYY8crNEVBwcHh8+bg0YwybJMVVUVq1atsj2+atUqW4rOwcHBwcHBwWF/c1Cl5ObNm8fdd99NVVUV1dXVvPHGG7S3t3PiiSd+0Yfm4ODg4ODgcAhzUAmmmTNn0tfXx1NPPUVXVxelpaX8/Oc//0TFWg4ODg4ODg4On5WDSjABnHzyyZx88slf9GE4ODg4ODg4HEYcNDVMDg4ODg4ODg5fFI5gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gcnBwcHBwcHDYCwfd8N19RZY/+ZI/zbaHGofr2p11H1446z68OFzXDYfv2j/Juj/puREsy7L29YAcHBwcHBwcHA5lnJRcAoLBIP/v//0/gsHgF30onzuH69qddTvrPhxw1n14rRsO37UfiHU7gikBlmWxfft2Dsfg2+G6dmfdzroPB5x1H17rhsN37Qdi3Y5gcnBwcHBwcHDYC45gcnBwcHBwcHDYC45gSoCiKJx33nkoivJFH8rnzuG6dmfdzroPB5x1H17rhsN37Qdi3U6XnIODg4ODg4PDXnAiTA4ODg4ODg4Oe8ERTA4ODg4ODg4Oe8ERTA4ODg4ODg4Oe8ERTA4ODg4ODg4Oe+HwHC4ziKeffppHH32U0047jcsuuyzpduvWreOBBx6gvr6ezMxMzjjjDE466aTP70D3M59k3WvXruWGG26Ie/zPf/4zxcXFB/gI9x+PP/44Tz75pO2x9PR0/vWvfyV9zaFwvT/tug+V6w3Q2dnJww8/zIoVK4hEIhQWFvKd73yHqqqqpK85FK75p133oXDNv/vd79LW1hb3+EknncTll1+e8DWHwrWGT7/2Q+F6AxiGwRNPPMF7771Hd3c3mZmZzJkzh3POOQdRTB4H2tfrflgLpi1btvDGG29QXl6+x+1aW1u57bbbOP744/n+97/Pxo0buffee0lLS+PII4/8nI52//FJ172bO++8E6/XG/t3WlragTq0A0ZpaSm//vWvY//e0x/VoXS9P826d3OwX+/+/n5+/etfM2bMGH7xi1+QlpZGS0uLbU1DORSu+WdZ924O5mt+2223YZpm7N91dXXcfPPNzJgxI+H2h8K13s2nXftuDubrDfDss8/y+uuv893vfpeSkhK2bdvG3/72N7xeL6eddlrC1+yP637YCqZQKMTdd9/NlVdeyfz58/e47WuvvUZOTk4sElNSUsLWrVt5/vnnD7o/sE+z7t2kp6fj8/kO8JEdWERRJCMj4xNteyhd70+z7t0c7Nf72WefJTs7m6uvvjr2WF5e3h5fcyhc88+y7t0czNd86M3+mWeeIT8/n5qamoTbHwrXejefdu27OZivN8CmTZuYOnUqkydPBqKf84ULF7J169akr9kf1/2wFUz33nsvkyZNYvz48XsVDps3b2b8+PG2xyZOnMhbb72FruvI8sFzGj/Nundz7bXXomkaJSUlnHPOOYwdO/YAH+X+p7m5mSuvvBJZlhkxYgRf/epXyc/PT7jtoXS9P826d3OwX+8lS5YwYcIE/vSnP7Fu3TqysrI46aSTOOGEE5K+5lC45p9l3bs52K/5bnRd57333mPu3LkIgpBwm0PhWifik6x9Nwf79R41ahSvv/46jY2NFBUVsWPHDjZu3Mill16a9DX747ofnJ+MfeT9999n+/bt3HbbbZ9o++7ubtLT022PpaenYxgGfX19ZGZmHojD3O982nVnZmby7W9/m6qqKnRd59133+Wmm27it7/97V5/wXyZGDFiBN/97ncpKiqiu7ub+fPn86tf/Yo//elPpKamxm1/qFzvT7vuQ+V6t7a28vrrrzN37lzOPvtstmzZwv3334+iKBxzzDEJX3MoXPPPsu5D5ZrvZtGiRfj9fubMmZN0m0PhWifik6z9ULneZ555JoFAgB//+MeIoohpmlx44YXMnj076Wv2x3U/7ARTe3s7//nPf/jlL3+Jqqqf+HVDFftug/S9KfkvC59l3UVFRRQVFcX+XV1dTXt7O88///xB9cc1adKk2P+XlZVRXV3N97//fd555x3mzZuX8DUH+/WGT7/uQ+V6m6bJsGHDuOiiiwCorKxk586dvPbaa0mFAxz81/yzrPtQuea7eeutt5g4cSJZWVl73O5gv9aJ+CRrP1Su9wcffMB7773HD37wA0pLS9mxYwf/+c9/YsXfydjX637YCaZt27bR09PDddddF3vMNE3Wr1/PK6+8wiOPPBJXGJuRkUF3d7ftsd7eXiRJIiUl5fM47H3ms6w7EdXV1bz33nsH8lAPOG63m7KyMpqamhI+fyhc70Tsbd2JOBivd2ZmJiUlJbbHSkpK+Pjjj5O+5lC45p9l3Yk4GK85QFtbG6tWreKaa67Z43aHwrUeyiddeyIOxuv98MMPc+aZZzJr1iwg+oOwra2NZ555Jqlg2h/X/bATTOPGjeOPf/yj7bG///3vFBUVceaZZyYUDSNGjGDp0qW2x1auXElVVdVBk+/+LOtOxPbt2z91EfGXDU3TaGhoYPTo0QmfPxSudyL2tu5EHIzXe+TIkTQ2Ntoea2xsJDc3N+lrDoVr/lnWnYiD8ZpDNMKSnp4eKwROxqFwrYfySdeeiIPxeofD4bh7liiK7Gk07v647oedcaXH46GsrMz2n8vlIjU1lbKyMgAeeeQR7rnnnthrTjrpJNrb22P+DQsWLGDBggWcfvrpX9QyPjWfZd0vvvgiixYtoqmpiZ07d/LII4/w8ccfc8opp3xRy/hMPPjgg6xbt47W1lY2b97MHXfcQTAYjKUpDsXrDZ9+3YfK9Z47dy6bN29m/vz5NDc3s3DhQt58801OPvnk2DaH4jX/LOs+VK65aZq8/fbbHHPMMUiSZHvuULzWg/k0az9UrveUKVOYP38+y5Yto7W1lUWLFvHCCy8wbdq02DYH4rofnHL6ANPV1UV7e3vs33l5efz85z/ngQce4NVXXyUzM5NvfOMbB10L6t4Yum5d13nooYfo7OxEVVVKS0u57rrrPtOvmC+Szs5O7rrrLnp7e0lLS2PEiBHccsstsV/eh+r1/rTrPlSu9/Dhw7nmmmt45JFHeOqpp8jLy+PSSy/lqKOOim1zKF7zz7LuQ+War169mvb2do499ti45w7Faz2YT7P2Q+V6f/Ob3+Sxxx7j3nvvpaenh6ysLE488UTOO++82DYH4roL1p5iWA4ODg4ODg4ODodfSs7BwcHBwcHB4dPiCCYHBwcHBwcHh73gCCYHBwcHBwcHh73gCCYHBwcHBwcHh73gCCYHBwcHBwcHh73gCCYHBwcHBwcHh73gCCYHBwcHBwcHh73gCCYHBwcHBwcHh73gCCYHh8OIt99+mwsuuCD234UXXshVV13F3/72Nzo7OwFYu3YtF1xwAR999FHC97jvvvu44IILPs/D/kRccMEF3HfffV/0YXzuzJ8/n0WLFn3Rh+HgcMjjjEZxcDgMufrqqykqKiISibB+/XqeeeYZ1q1bFzeg2eHLz9NPP82RRx7J9OnTv+hDcXA4pHEEk4PDYUhpaSnDhg0DYOzYsZimyVNPPcXixYvJysr6wo6rq6sLl8uF1+v9wo7hUCESidDb20tOTs4XfSgODocEjmBycHBgxIgRALS1tX3ugqm/v59FbyiKcAAACHZJREFUixbx/vvvs2bNGm6//XYqKioA+Otf/8q6dev461//anvN448/zpNPPsnjjz8e936vv/46L7zwAm1tbeTn53Peeecxa9Ys2zbd3d08/vjjLFu2LDa8c86cOZxzzjmxie+tra1873vf45JLLkEURV5++WV6e3spKyvj0ksvpbq62vaeW7du5cknn2TDhg1EIhGKi4s566yzmDlzJgA7duzg2muv5aqrruK4446zvXb58uXcdtttXHvttUydOjW2vjvuuIOnnnqK5cuXo6oqkyZN4rLLLosJyt2p0XfeeYd33nkHgJqaGq6//nq6u7v5/ve/z6hRo5g1axZHHnkkaWlpn+USOTg44AgmBwcHoLm5GcB2QzVNE8Mw4rbdH/O6w+EwS5cuZeHChaxYsQJRFJk0aRI//vGPKSkp+czvu2TJklgNlsvl4rXXXuOuu+5CkqTYVPLu7m5+/vOfI4oi5513Hvn5+WzatIn58+fT1tbG1VdfbXvPV199leLiYi677DIAHnvsMW677Tb++te/xoTLmjVruPXWWxkxYgRXXHEFXq+XDz74gDvvvJNIJMKcOXOoqKigsrKSt956K04wvf3226SnpzNp0iTb43fccQczZ87kuOOOo66ujkcffRQgdow333wzN954I2PGjOHcc88FiB1TTk4O11xzDe+//z4PPfQQ999/P+PHj2fWrFlMmzYNj8fzmc+zg8PhiCOYHBwOQ3aLIU3TWLduHfPnz8fj8TB16lQaGhoAuPPOO/frPnVdZ9WqVSxcuJDFixdjGAbj/3979xfSdBfHcfztbNpKy6nM8t9uyv6p0+WYDvNPNxGRBF2kRWCgF1JEFIgUBQXlnSF4UyIWi6ywKCG6VMQ/BaaU4kUQ4YUSlJZTRzbZ81yIoz2O9mR/Hp76vG7G78f5nbPfYOzD95zfWXY2NTU1OBwOVq9e/d1jzMzMUF9fT1xcHAB2u50zZ85w+/btQGC6d+8ec3NzNDQ0BKarsrKyiIqKwu12U1ZWFhTaTCYTdXV1GAyLz8iYzWbOnj3L0NBQoHLV0tJCWloaFy5cCFSocnJy8Hg8tLW1UVRUhMFgoKSkhNbWViYmJkhOTgYWK2wDAwPs2bMncO2S3bt3U1ZWBkB2djZv376ls7OTmpoaIiIiyMjIICIignXr1i2reBkMBhwOBw6HIxBQ+/r6uHbtGtevX8dut1NYWEhubi5Go/G7P3uR350Ck8gf6Ny5c0HH6enpVFVVERcXFwhMR44cITMzc9m1HR0d9Pf3f9N4Y2NjXLx4Ea/XS1ZWFseOHcPpdLJ27dqV30QImZmZgbAEi6GhoKCA9vZ2JicnSUhIYHBwkB07dmA2m4MqaLm5ubjdbkZHR4MCk91uD4QlAKvVCixOX8JidW58fJyjR48CBPVpt9sZHBxkYmKC1NRUdu3axa1bt+jq6uLw4cMA9Pb24vP5KC0tXXY/eXl5QcdWqxWfz8f09HTQfYYTHR2Ny+XC5XLh9XoZGBigt7eXq1evEh0dTV1dHVu3bv3X/Yn8iRSYRP5AJ06cICUlhcjISNavX4/ZbF7WJikpKbAw/EsrWQcTGRnJmjVrmJ2dxev14vV6+fTp0w8PTKFCxNK5mZkZEhISmJ6e5vnz51RUVITsw+PxBB3HxMQEHS9VYz5//gwsTvEBuN1u3G53yD5nZmYCfe3cuZPu7m7Ky8sxGAx0dXWxadMm0tLSll0XbuyVmJ+fZ25uDq/Xi9/vx2QysWqVfgpEwtG3ROQPlJKSEjIM/Sypqak0NTXx6tUrenp6ePToEW63my1btuByucjPzw8ZdoxGIz6fb9n5pQDyT0vhJdS52NjYwKvVaqW8vDxkH6HC49csBcgDBw7gdDpDtlmafgMoLS3l6dOnvHz5ksTERF6/fk1VVdU3jfmtPB4Pz549o6+vj9HRUWJiYnA6nVRUVLBt2zYiIiJ+6vgivwMFJhH5ZTIyMsjIyKCyspLh4WF6enpoa2ujtbWV7du343K5KC4uJioqCgCLxcL09DQfP34MBKqFhQVevHgRsv+RkZGgtn6/n/7+fpKSkkhISAAWp8mGhoZISkpaVsFZieTkZDZu3MjY2Fhgmu1rbDYb8fHxdHZ2kpiYiNFopLCwcMXjG43GkBWnhYUFuru76e/vZ2RkhFWrVpGXl0dtbS02m01VJZFvpG+MiPxyBoMBm82GzWajurqawcFBenp6uHHjBps3bw5sK+Byubh79y6NjY3s378fn8/HkydP8Pv9IfuNjY3l0qVLHDx4MPCU3Pj4OKdOnQq0OXToEMPDw5w/f569e/cGNvB89+4dQ0NDVFdXB8LVv1VdXU19fT2XL1+muLiY+Ph4ZmdnGR8f582bN5w+fTro3ouKinj8+DEmkwmn0/ld+06lp6czOjrKwMAAZrMZk8lEcnIyU1NTNDc3Y7PZOH78OA6Hg+jo6BWPI/KnU2ASkf9UVFQU+fn55Ofn4/V6gxZYWywWamtraWtro6GhAbPZzL59+/B4PLS3ty/rKy8vj7S0NO7cucP79+/ZsGEDJ0+eDOyFBItTbvX19dy/f5+Ojg4mJycxmUxYLBZycnJWtK4qMzOTK1eu8ODBA27evMns7CyxsbGkpqZSUFCwrH1paSkPHz7E5/NRUlLyzeN9qbKykpaWFhobG5mfnw/sw2Q2m2lubv4hVTQRgYi/fsSmKiIiIiK/Mf35roiIiEgYmpITkRXz+/1hd/7+52aMIiL/RwpMIrJi7e3tIdcSfampqQmLxfKL3pGIyM+hNUwismJTU1N8+PDhq22sVqseYReR/z0FJhEREZEwtOhbREREJAwFJhEREZEwFJhEREREwlBgEhEREQlDgUlEREQkDAUmERERkTAUmERERETC+Bu0x5HKjB2L5AAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot just the 'new' samples\n", "data_new = data.query(\"period == 'new'\")\n", "fig, ax = plt.subplots(figsize=(6, 6))\n", "sn.scatterplot(\n", " data=data_new,\n", " x=\"PH_\",\n", " y=\"CA_mg/l\",\n", " hue=\"pred\",\n", " ax=ax,\n", " hue_order=[\"outlier\", \"inlier\"],\n", ")\n", "_ = ax.set_title(\"'New' data only\")\n", "plt.tight_layout()\n", "png_path = os.path.join(fold_path, \"isoforest_ca_ph_plot.png\")\n", "plt.savefig(png_path, dpi=200)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 4.2. `CA`, `PH`, `ILAL` and `RAL`" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/opt/conda/lib/python3.10/site-packages/sklearn/base.py:450: UserWarning: X does not have valid feature names, but IsolationForest was fitted with feature names\n", " warnings.warn(\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "The total number of samples in the dataset is: 11439.\n", "\n", "The total number of outliers detected is 115:\n", " 100 in the 'historic' period\n", " 15 in the 'new' period\n", "\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
vannmiljo_codesample_datelabperioddepth1depth2CA_mg/lPH_<ubenevnt>ILAL_µg/l AlRAL_µg/l Alpred
1620019-1010222022-09-06Eurofinsnew0.00.01.405.7130.0180.0outlier
2205019-791482022-09-06Eurofinsnew0.00.01.705.8150.0190.0outlier
4073021-292462022-09-30Eurofinsnew0.00.01.205.1140.0240.0outlier
5752022-320182022-09-23Eurofinsnew0.00.09.607.59.711.0outlier
5846022-320192022-07-22Eurofinsnew0.00.04.606.95.15.9outlier
5847022-320192022-08-01Eurofinsnew0.00.05.407.05.49.2outlier
5849022-320192022-09-23Eurofinsnew0.00.07.406.917.021.0outlier
5941022-320202022-07-22Eurofinsnew0.00.06.007.06.26.7outlier
6763022-589042022-07-22Eurofinsnew0.00.05.706.85.05.0outlier
8335024-588942022-09-07Eurofinsnew0.00.04.206.95.05.0outlier
14226027-792782022-07-05Eurofinsnew0.00.04.807.15.47.3outlier
14227027-792782022-08-02Eurofinsnew0.00.04.907.25.06.9outlier
14228027-792782022-09-06Eurofinsnew0.00.04.607.05.05.0outlier
19873036-587522022-08-23Eurofinsnew0.00.00.645.3140.0170.0outlier
34645082-588742022-07-05Eurofinsnew0.00.00.545.6130.0170.0outlier
\n", "
" ], "text/plain": [ " vannmiljo_code sample_date lab period depth1 depth2 CA_mg/l \\\n", "1620 019-101022 2022-09-06 Eurofins new 0.0 0.0 1.40 \n", "2205 019-79148 2022-09-06 Eurofins new 0.0 0.0 1.70 \n", "4073 021-29246 2022-09-30 Eurofins new 0.0 0.0 1.20 \n", "5752 022-32018 2022-09-23 Eurofins new 0.0 0.0 9.60 \n", "5846 022-32019 2022-07-22 Eurofins new 0.0 0.0 4.60 \n", "5847 022-32019 2022-08-01 Eurofins new 0.0 0.0 5.40 \n", "5849 022-32019 2022-09-23 Eurofins new 0.0 0.0 7.40 \n", "5941 022-32020 2022-07-22 Eurofins new 0.0 0.0 6.00 \n", "6763 022-58904 2022-07-22 Eurofins new 0.0 0.0 5.70 \n", "8335 024-58894 2022-09-07 Eurofins new 0.0 0.0 4.20 \n", "14226 027-79278 2022-07-05 Eurofins new 0.0 0.0 4.80 \n", "14227 027-79278 2022-08-02 Eurofins new 0.0 0.0 4.90 \n", "14228 027-79278 2022-09-06 Eurofins new 0.0 0.0 4.60 \n", "19873 036-58752 2022-08-23 Eurofins new 0.0 0.0 0.64 \n", "34645 082-58874 2022-07-05 Eurofins new 0.0 0.0 0.54 \n", "\n", " PH_ ILAL_µg/l Al RAL_µg/l Al pred \n", "1620 5.7 130.0 180.0 outlier \n", "2205 5.8 150.0 190.0 outlier \n", "4073 5.1 140.0 240.0 outlier \n", "5752 7.5 9.7 11.0 outlier \n", "5846 6.9 5.1 5.9 outlier \n", "5847 7.0 5.4 9.2 outlier \n", "5849 6.9 17.0 21.0 outlier \n", "5941 7.0 6.2 6.7 outlier \n", "6763 6.8 5.0 5.0 outlier \n", "8335 6.9 5.0 5.0 outlier \n", "14226 7.1 5.4 7.3 outlier \n", "14227 7.2 5.0 6.9 outlier \n", "14228 7.0 5.0 5.0 outlier \n", "19873 5.3 140.0 170.0 outlier \n", "34645 5.6 130.0 170.0 outlier " ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Columns of interest\n", "key_cols = [\"vannmiljo_code\", \"sample_date\", \"lab\", \"period\", \"depth1\", \"depth2\"]\n", "par_cols = [\"CA_mg/l\", \"PH_\", \"ILAL_µg/l Al\", \"RAL_µg/l Al\"]\n", "\n", "# Run algorithm\n", "data = df[key_cols + par_cols].dropna()\n", "data = utils.isolation_forest(data, par_cols, contamination=0.01)\n", "\n", "# Summarise results\n", "all_out = data.query(\"pred == 'outlier'\")\n", "his_out = data.query(\"(pred == 'outlier') and (period == 'historic')\")\n", "new_out = data.query(\"(pred == 'outlier') and (period == 'new')\")\n", "\n", "print(f\"The total number of samples in the dataset is: {len(data)}.\\n\")\n", "print(\n", " f\"The total number of outliers detected is {len(all_out)}:\\n\"\n", " f\" {len(his_out)} in the 'historic' period\\n\"\n", " f\" {len(new_out)} in the 'new' period\\n\"\n", ")\n", "\n", "new_out" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 4.3. All parameters *except* `LAL` and `TEMP`" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "/opt/conda/lib/python3.10/site-packages/sklearn/base.py:450: UserWarning: X does not have valid feature names, but IsolationForest was fitted with feature names\n", " warnings.warn(\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "The total number of samples in the dataset is: 1759.\n", "\n", "The total number of outliers detected is 18:\n", " 5 in the 'historic' period\n", " 13 in the 'new' period\n", "\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
vannmiljo_codesample_datelabperioddepth1depth2ALK_mmol/lANC_µekv/lCA_mg/lCL_mg/l...N-NO3_µg/l NN-TOT_µg/l NNA_mg/lP-TOT_µg/l PPH_<ubenevnt>RAL_µg/l AlSIO2_µg/l SiSO4_mg/lTOC_mg/l Cpred
3930021-284502022-07-08Eurofinsnew0.00.00.1579.07.735.0...270.0530.016.014.06.711.01309.1211724.664.6outlier
3932021-284502022-08-09Eurofinsnew0.00.00.0969.05.026.0...270.03600.013.022.06.631.01215.6125173.907.0outlier
3935021-284502022-09-09Eurofinsnew0.00.00.16110.08.633.0...720.0830.015.08.46.69.11402.6298274.824.4outlier
6879022-970112022-07-06Eurofinsnew0.00.00.25270.04.26.6...150.0290.04.97.07.127.02477.9793612.474.9outlier
6881022-970112022-08-03Eurofinsnew0.00.00.25270.04.17.1...160.0270.05.55.77.028.02805.2596543.064.6outlier
6884022-970112022-09-07Eurofinsnew0.00.00.24230.03.97.5...140.0300.04.94.17.029.02805.2596542.675.1outlier
8333024-588942022-07-06Eurofinsnew0.00.00.15-0.33.813.0...810.0990.05.312.06.77.3701.3149133.983.0outlier
8334024-588942022-08-02Eurofinsnew0.00.00.16120.04.111.0...1100.01200.06.66.97.08.91122.1038623.932.5outlier
8335024-588942022-09-07Eurofinsnew0.00.00.1697.04.211.0...1300.01600.06.64.86.95.01122.1038624.662.1outlier
14226027-792782022-07-05Eurofinsnew0.00.00.30230.04.818.0...530.0740.09.637.07.17.31683.1557923.673.6outlier
14227027-792782022-08-02Eurofinsnew0.00.00.30230.04.918.0...810.0950.010.031.07.26.92197.4533953.873.3outlier
14228027-792782022-09-06Eurofinsnew0.00.00.27240.04.617.0...590.0790.09.333.07.05.01215.6125173.174.1outlier
17198030-588382022-09-06Eurofinsnew0.00.00.11100.03.64.7...1700.02000.02.6110.06.29.1794.8235691.942.8outlier
\n", "

13 rows × 24 columns

\n", "
" ], "text/plain": [ " vannmiljo_code sample_date lab period depth1 depth2 ALK_mmol/l \\\n", "3930 021-28450 2022-07-08 Eurofins new 0.0 0.0 0.15 \n", "3932 021-28450 2022-08-09 Eurofins new 0.0 0.0 0.09 \n", "3935 021-28450 2022-09-09 Eurofins new 0.0 0.0 0.16 \n", "6879 022-97011 2022-07-06 Eurofins new 0.0 0.0 0.25 \n", "6881 022-97011 2022-08-03 Eurofins new 0.0 0.0 0.25 \n", "6884 022-97011 2022-09-07 Eurofins new 0.0 0.0 0.24 \n", "8333 024-58894 2022-07-06 Eurofins new 0.0 0.0 0.15 \n", "8334 024-58894 2022-08-02 Eurofins new 0.0 0.0 0.16 \n", "8335 024-58894 2022-09-07 Eurofins new 0.0 0.0 0.16 \n", "14226 027-79278 2022-07-05 Eurofins new 0.0 0.0 0.30 \n", "14227 027-79278 2022-08-02 Eurofins new 0.0 0.0 0.30 \n", "14228 027-79278 2022-09-06 Eurofins new 0.0 0.0 0.27 \n", "17198 030-58838 2022-09-06 Eurofins new 0.0 0.0 0.11 \n", "\n", " ANC_µekv/l CA_mg/l CL_mg/l ... N-NO3_µg/l N N-TOT_µg/l N NA_mg/l \\\n", "3930 79.0 7.7 35.0 ... 270.0 530.0 16.0 \n", "3932 69.0 5.0 26.0 ... 270.0 3600.0 13.0 \n", "3935 110.0 8.6 33.0 ... 720.0 830.0 15.0 \n", "6879 270.0 4.2 6.6 ... 150.0 290.0 4.9 \n", "6881 270.0 4.1 7.1 ... 160.0 270.0 5.5 \n", "6884 230.0 3.9 7.5 ... 140.0 300.0 4.9 \n", "8333 -0.3 3.8 13.0 ... 810.0 990.0 5.3 \n", "8334 120.0 4.1 11.0 ... 1100.0 1200.0 6.6 \n", "8335 97.0 4.2 11.0 ... 1300.0 1600.0 6.6 \n", "14226 230.0 4.8 18.0 ... 530.0 740.0 9.6 \n", "14227 230.0 4.9 18.0 ... 810.0 950.0 10.0 \n", "14228 240.0 4.6 17.0 ... 590.0 790.0 9.3 \n", "17198 100.0 3.6 4.7 ... 1700.0 2000.0 2.6 \n", "\n", " P-TOT_µg/l P PH_ RAL_µg/l Al SIO2_µg/l Si SO4_mg/l \\\n", "3930 14.0 6.7 11.0 1309.121172 4.66 \n", "3932 22.0 6.6 31.0 1215.612517 3.90 \n", "3935 8.4 6.6 9.1 1402.629827 4.82 \n", "6879 7.0 7.1 27.0 2477.979361 2.47 \n", "6881 5.7 7.0 28.0 2805.259654 3.06 \n", "6884 4.1 7.0 29.0 2805.259654 2.67 \n", "8333 12.0 6.7 7.3 701.314913 3.98 \n", "8334 6.9 7.0 8.9 1122.103862 3.93 \n", "8335 4.8 6.9 5.0 1122.103862 4.66 \n", "14226 37.0 7.1 7.3 1683.155792 3.67 \n", "14227 31.0 7.2 6.9 2197.453395 3.87 \n", "14228 33.0 7.0 5.0 1215.612517 3.17 \n", "17198 110.0 6.2 9.1 794.823569 1.94 \n", "\n", " TOC_mg/l C pred \n", "3930 4.6 outlier \n", "3932 7.0 outlier \n", "3935 4.4 outlier \n", "6879 4.9 outlier \n", "6881 4.6 outlier \n", "6884 5.1 outlier \n", "8333 3.0 outlier \n", "8334 2.5 outlier \n", "8335 2.1 outlier \n", "14226 3.6 outlier \n", "14227 3.3 outlier \n", "14228 4.1 outlier \n", "17198 2.8 outlier \n", "\n", "[13 rows x 24 columns]" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Columns of interest\n", "key_cols = [\"vannmiljo_code\", \"sample_date\", \"lab\", \"period\", \"depth1\", \"depth2\"]\n", "excl_cols = [\"TEMP_°C\", \"LAL_µg/l Al\"]\n", "par_cols = [col for col in df.columns if col not in (key_cols + excl_cols)]\n", "\n", "# Run algorithm\n", "data = df[key_cols + par_cols].dropna()\n", "data = utils.isolation_forest(data, par_cols, contamination=0.01)\n", "\n", "# Summarise results\n", "all_out = data.query(\"pred == 'outlier'\")\n", "his_out = data.query(\"(pred == 'outlier') and (period == 'historic')\")\n", "new_out = data.query(\"(pred == 'outlier') and (period == 'new')\")\n", "\n", "print(f\"The total number of samples in the dataset is: {len(data)}.\\n\")\n", "print(\n", " f\"The total number of outliers detected is {len(all_out)}:\\n\"\n", " f\" {len(his_out)} in the 'historic' period\\n\"\n", " f\" {len(new_out)} in the 'new' period\\n\"\n", ")\n", "\n", "new_out" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.6" } }, "nbformat": 4, "nbformat_minor": 4 }