{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "## Predictive Analytics: Choosing the model with highest ROI for Term Deposit Campaigns\n", "\n", "* In this analysis I am trying to fit a predictive model to predict whether a bank customer will subscribe to a term deposit and then try to report the results in a clear and concise manner using the package 'modelplotpy'.\n", "* We will also look at how the predictive model is doing locally using the package 'lime'.\n", "* The dataset has various kinds of features associated with 1 row ( = 1 customer). Lets explore.\n", "\n", "\n", "\n", "\n", "* Importing required libraries" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\cgokh\\Anaconda3\\lib\\site-packages\\sklearn\\ensemble\\weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n", " from numpy.core.umath_tests import inner1d\n" ] } ], "source": [ "# Importing libraries\n", "import pandas as pd\n", "from sklearn.model_selection import train_test_split\n", "from sklearn.ensemble import RandomForestClassifier\n", "from sklearn.ensemble import GradientBoostingClassifier\n", "from sklearn.metrics import recall_score, accuracy_score, precision_score, confusion_matrix\n", "import seaborn as sns\n", "import modelplotpy as mp\n", "import numpy as np\n", "import warnings\n", "warnings.filterwarnings(\"ignore\")" ] }, { "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", "
agejobmaritaleducationdefaulthousingloancontactmonthday_of_week...campaignpdayspreviouspoutcomeemp.var.ratecons.price.idxcons.conf.idxeuribor3mnr.employedy
056housemaidmarriedbasic.4ynononotelephonemaymon...19990nonexistent1.193.994-36.44.8575191.0no
157servicesmarriedhigh.schoolunknownnonotelephonemaymon...19990nonexistent1.193.994-36.44.8575191.0no
237servicesmarriedhigh.schoolnoyesnotelephonemaymon...19990nonexistent1.193.994-36.44.8575191.0no
340admin.marriedbasic.6ynononotelephonemaymon...19990nonexistent1.193.994-36.44.8575191.0no
456servicesmarriedhigh.schoolnonoyestelephonemaymon...19990nonexistent1.193.994-36.44.8575191.0no
\n", "

5 rows × 21 columns

\n", "
" ], "text/plain": [ " age job marital education default housing loan contact \\\n", "0 56 housemaid married basic.4y no no no telephone \n", "1 57 services married high.school unknown no no telephone \n", "2 37 services married high.school no yes no telephone \n", "3 40 admin. married basic.6y no no no telephone \n", "4 56 services married high.school no no yes telephone \n", "\n", " month day_of_week ... campaign pdays previous poutcome emp.var.rate \\\n", "0 may mon ... 1 999 0 nonexistent 1.1 \n", "1 may mon ... 1 999 0 nonexistent 1.1 \n", "2 may mon ... 1 999 0 nonexistent 1.1 \n", "3 may mon ... 1 999 0 nonexistent 1.1 \n", "4 may mon ... 1 999 0 nonexistent 1.1 \n", "\n", " cons.price.idx cons.conf.idx euribor3m nr.employed y \n", "0 93.994 -36.4 4.857 5191.0 no \n", "1 93.994 -36.4 4.857 5191.0 no \n", "2 93.994 -36.4 4.857 5191.0 no \n", "3 93.994 -36.4 4.857 5191.0 no \n", "4 93.994 -36.4 4.857 5191.0 no \n", "\n", "[5 rows x 21 columns]" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data = pd.read_csv(\"../bank-additional/bank-additional-full.csv\", sep=\";\")\n", "data.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* We are using the UCI [Bank Marketing dataset](https://archive.ics.uci.edu/ml/datasets/bank+marketing)\n", "* As we can see above, each row represents a customer of the bank with details related to the customer in subsequent columns.\n", "* We have demographic details, product (bank products) related details, marketing campaign details and some economic factors.\n", "* The target variable for this study is whether the customer subscribed to term deposit or not?\n", "* 'duration' feature is marked as a potential leak by the data creators so we will not use this feature in the analysis. With 'duration' any model performs very well, without it it will be interesting to see how results shape up. That is good for us because in real life we seldom find 'wizard models' (models that perform exceptionally well)\n", "\n", "
\n", "\n", "#### Data preparation and fitting predictive models using sklearn" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "GradientBoostingClassifier(criterion='friedman_mse', init=None,\n", " learning_rate=0.1, loss='deviance', max_depth=10,\n", " max_features=None, max_leaf_nodes=None,\n", " min_impurity_decrease=0.0, min_impurity_split=None,\n", " min_samples_leaf=1, min_samples_split=2,\n", " min_weight_fraction_leaf=0.0, n_estimators=500,\n", " presort='auto', random_state=None, subsample=0.85, verbose=0,\n", " warm_start=False)" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Firstly, lets change the name of the target column as well as the 'yes' level to a more meaningful name\n", "data['y'].replace({'yes': 'term_deposit'}, inplace=True)\n", "\n", "data.drop(['month', 'day_of_week', 'duration'], axis=1, inplace=True)\n", "\n", "y = data['y']\n", "#encode_y = LabelEncoder()\n", "#y_encoded = encode_y.fit_transform(y)\n", "\n", "data.drop('y', axis=1, inplace=True)\n", "\n", "data = pd.get_dummies(data)\n", "\n", "X_train, X_test, y_train, y_test = train_test_split(data, y, test_size=0.1, random_state=123)\n", "\n", "rf_classifier = RandomForestClassifier(n_estimators=500, criterion='gini', max_depth=5, min_samples_split=0.1,\n", " max_features='auto')\n", "rf_classifier.fit(X_train, y_train)\n", "\n", "gbm_classifier = GradientBoostingClassifier(learning_rate=0.1, n_estimators=500, subsample=0.85, max_depth=10)\n", "gbm_classifier.fit(X_train, y_train)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "#### Checking the model performance using traditional metrics" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "gbm_preds = gbm_classifier.predict(X_test)\n", "rf_preds = rf_classifier.predict(X_test)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "#### Random Forest Confusion Matrix (Threshold= 0.5)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[3666, 5],\n", " [ 439, 9]], dtype=int64)" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "confusion_matrix(y_test, rf_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* So confusion matrix for Random Forest is interesting.\n", "* At 0.5 threshold which is the default threshold for rf.predict() method, we have just 17 observations predicted to be subscribing for TERM DEPOSIT.\n", "* Lets look at the distribution of the probabilities" ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "rf_probs = rf_classifier.predict_proba(X_test)\n", "rf_probs = np.array([i[1] for i in rf_probs])\n", "sns.distplot(rf_probs, kde=True, color='r')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* We can see that there are hardly any probabilities predicted beyond 0.5\n", "* This is why 0.5 threshold is yielding only 17 predicted 'term_deposit' classes\n", "* What to make of this model? Is it good? Is it bad?\n", "* We will get to that later.\n", "* First let us compare this to GBM model which we have fit" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[3516, 155],\n", " [ 309, 139]], dtype=int64)" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "confusion_matrix(y_test, gbm_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* The confusion matrix from gbm predictions shows a completely opposite picture.\n", "* Lets check the distribution of probabilities." ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "gbm_probs = gbm_classifier.predict_proba(X_test)\n", "gbm_probs = np.array([i[1] for i in gbm_probs])\n", "sns.distplot(gbm_probs, kde=True, color='b')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* We can see here that GBM model predicts more observations having probabilities > 0.5 and so we have larger numbers in that right column. By the way, the Confusion Matrix has Predictions on horizontal axis and Actual values on vertical axis.\n", "* Again, we have confusion matrices of both models. We can easily get precision and recall scores too but how to know which model is better?\n", "\n", "
\n", "\n", "* Lets try to use an additional tool to judge which model is better.\n", "* Package **modelplotpy** provides us with 4 graphs. 1 of them is 'Cummulative Gains'.\n", "* Lets get the graph for RF model. I will mention what the graph shows below it." ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Default scope value no_comparison selected, single evaluation line will be plotted.\n", "The label with smallest class is term_deposit\n", "Target class term_deposit, dataset test data and model random forest.\n", "When we select 20% with the highest probability according to model random forest, this selection holds 59% of all term_deposit cases in dataset test data.\n", "The cumulative gains plot is saved in C:\\Github\\JupyterNotebooks\\Notebooks/Cumulative gains plot.png\n" ] }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "obj = mp.modelplotpy(feature_data = [X_train, X_test]\n", " , label_data = [y_train, y_test]\n", " , dataset_labels = ['train data', 'test data']\n", " , models = [rf_classifier, gbm_classifier]\n", " , model_labels = ['random forest', 'gradient boosting']\n", " )\n", "\n", "# transform data generated with prepare_scores_and_deciles into aggregated data for chosen plotting scope \n", "ps = obj.plotting_scope(select_model_label = ['random forest'], select_dataset_label = ['test data'])\n", "\n", "mp.plot_cumgains(ps, highlight_decile=2)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* The above graph has inverted deciles of predicted probability distribution on X-axis\n", "* So 1 is the 9th decile, 2 is the 8th decible and so on.\n", "* Y-axis has % of actual positive classes we retain if we use the particular X-axis decile as our threshold.\n", "* So for example, if we use Decile 2 (which is actually 8th Decile) as our threshold for classifying term deposits, we will get 58% of the total actual term deposit subscribing customers in the test set.\n", "* So in our case we have 448 total Positive classes in our test set. Out of that, if we use 8th decile of the current Predicted Probability distribution as our threshold, we will correctly TARGET 265 of those 448 customers.\n", "* One thing this graph does not tell us is how many False Positives will we have in the Top 20% of those probabilities.\n", "* Let us find that out ourselves using the following code." ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 0, 559],\n", " [ 0, 265]], dtype=int64)" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Extracting indices of Probs. over 80 percentile or 8th Decilbe (Top 20% probabilities)\n", "rf_probs_top20percent_indices = np.where(rf_probs > np.quantile(rf_probs, 0.8))\n", "\n", "# Using those indices to get the true labels of those probs.\n", "rf_probs_top20percent_y_test = np.array(y_test)[rf_probs_top20percent_indices]\n", "\n", "# Separating top 20% probs in a separate array\n", "rf_probs_top20percent_probs = rf_probs[rf_probs_top20percent_indices]\n", "\n", "# creating all positive predictions for top 20% probabilities\n", "rf_probs_top20percent_preds = np.where(rf_probs_top20percent_probs > 0, 'term_deposit', 'no')\n", "\n", "confusion_matrix(rf_probs_top20percent_y_test, rf_probs_top20percent_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* We can see we got 265 True Positives (~59% of 448 total positive cases retained).\n", "* And as a result of using top 20% we will end up campaigning on 559 customers wrongly.\n", "* Suppose our earnings per positive customer is 25 USD and cost per wrong customer is 5 USD (I am just making up the numbers here), then our Net profit if we use this model will be (265 x 25)-(559 x 5) = 6625 - 2795 = 3830 USD.\n", "* Let us do this same exercise for GBM now and see how much net profit that model will have?" ] }, { "cell_type": "code", "execution_count": 13, "metadata": { "scrolled": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Default scope value no_comparison selected, single evaluation line will be plotted.\n", "The label with smallest class is term_deposit\n", "Target class term_deposit, dataset test data and model gradient boosting.\n", "When we select 30% with the highest probability according to model gradient boosting, this selection holds 62% of all term_deposit cases in dataset test data.\n", "The cumulative gains plot is saved in C:\\Github\\JupyterNotebooks\\Notebooks/Cumulative gains plot.png\n" ] }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# transform data generated with prepare_scores_and_deciles into aggregated data for chosen plotting scope \n", "ps_gbm = obj.plotting_scope(select_model_label = ['gradient boosting'], select_dataset_label = ['test data'])\n", "\n", "mp.plot_cumgains(ps_gbm, highlight_decile=3)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* 'Cummulative gains' plot for GBM model shows us that top 30% probs. recover 60% positive cases from the test set.\n", "* Lets calculate the net profit." ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 0, 956],\n", " [ 0, 280]], dtype=int64)" ] }, "execution_count": 14, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Extracting indices of Probs. over 80 percentile or 8th Decilbe (Top 20% probabilities)\n", "gbm_probs_top20percent_indices = np.where(gbm_probs > np.quantile(gbm_probs, 0.7))\n", "\n", "# Using those indices to get the true labels of those probs.\n", "gbm_probs_top20percent_y_test = np.array(y_test)[gbm_probs_top20percent_indices]\n", "\n", "# Separating top 20% probs in a separate array\n", "gbm_probs_top20percent_probs = gbm_probs[gbm_probs_top20percent_indices]\n", "\n", "# creating all positive predictions for top 20% probabilities\n", "gbm_probs_top20percent_preds = np.where(gbm_probs_top20percent_probs > 0, 'term_deposit', 'no')\n", "\n", "confusion_matrix(gbm_probs_top20percent_y_test, gbm_probs_top20percent_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* So for GBM the net profit is (273 x 25) - (963 x 5) = 2010 USD.\n", "* So we can see that if we use GBM model we will have a reduction of 3830 - 2010 = 1820 USD in profits.\n", "* Hence, we should use Random Forest model for our marketing efforts.\n", "\n", "\n", "## Will an optimized f-score threshold give us a better ROI?\n", "\n", "* Right now we have used deciles to decide the threshold and then calculated ROI.\n", "* Can it happen that if we choose some other threshold, we may get better ROI?\n", "* Lets optimize the threshold such that we get the best f-score and calculate ROI using that.\n", "* So that we can compare that and this." ] }, { "cell_type": "code", "execution_count": 38, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0,0.5,'%')" ] }, "execution_count": 38, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "from sklearn.metrics import precision_recall_curve\n", "from sklearn.preprocessing import LabelEncoder\n", "import matplotlib.pyplot as plt\n", "\n", "%matplotlib inline\n", "\n", "lbl_enc = LabelEncoder()\n", "y_test_encoded = lbl_enc.fit_transform(y_test)\n", "\n", "precision, recall, thresholds = precision_recall_curve(y_test_encoded, rf_probs, pos_label=1)\n", "\n", "fig=plt.figure(figsize=(10, 8), dpi= 80, facecolor='w', edgecolor='k')\n", "thresholds = np.append(thresholds, 1)\n", "queue_rate = [] \n", "for threshold in thresholds:\n", " queue_rate.append((rf_probs >= threshold).mean())\n", "plt.plot(thresholds, precision, color=sns.color_palette()[0])\n", "plt.plot(thresholds, recall, color=sns.color_palette()[1]) \n", "plt.plot(thresholds, queue_rate, color=sns.color_palette()[2])\n", "leg = plt.legend(('precision', 'recall', 'queue_rate'), frameon=True)\n", "leg.get_frame().set_edgecolor('k')\n", "plt.xlabel('threshold')\n", "plt.ylabel('%')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* Above code is the [courtesy of the following post](https://blog.insightdatascience.com/visualizing-machine-learning-thresholds-to-make-better-business-decisions-4ab07f823415/)\n", "* This plot shows us 3 things\n", " 1. Queue Rate - Out of all the score test observations, % of users selected to taken action upon (marketing in our case. Can be churn etc.. in other cases)\n", " 2. Precision - Of all the customers to whom we do end up marketing, % of customers who will actually but term deposit.\n", " 3. Recall - Of all the customers who would have bought term deposit, % of customers to whom we end up marketing.\n", "* Second and third ones we already know.\n", "* So for our Random Forest model, the best threshold for an optimum precision and recall seems to be around 0.3\n", "* Lets see how our ROI will look like if we select 0.3" ] }, { "cell_type": "code", "execution_count": 52, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 0, 259],\n", " [ 0, 191]], dtype=int64)" ] }, "execution_count": 52, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Extracting indices of Probs. over 80 percentile or 8th Decilbe (Top 20% probabilities)\n", "rf_probs_top20percent_indices = np.where(rf_probs > 0.3)\n", "\n", "# Using those indices to get the true labels of those probs.\n", "rf_probs_top20percent_y_test = np.array(y_test)[rf_probs_top20percent_indices]\n", "\n", "# Separating top 20% probs in a separate array\n", "rf_probs_top20percent_probs = rf_probs[rf_probs_top20percent_indices]\n", "\n", "# creating all positive predictions for top 20% probabilities\n", "rf_probs_top20percent_preds = np.where(rf_probs_top20percent_probs > 0, 'term_deposit', 'no')\n", "\n", "confusion_matrix(rf_probs_top20percent_y_test, rf_probs_top20percent_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* Profit = Earnings - Costs = 4775 - 1295 = USD 3480\n", "* This is lower than ~3800 if we select the threshold by Best deciles method.\n", "* One thing to also note here is that the Avg. cost to take marketing action (USD 5 taken by us) and Avg. earning per marketed customer (USD 25 taken by us) hugely alter the way the PROFIT figure shapes up.\n", "* So it may happen that if these numbers are different in real life, we might have to make decisions accordingly.\n", "* But as of now, according to whatever figures we have taken, Method 1 has proved to be better.\n", "\n", "### Examine cases for different values of costs and gains\n", "\n", "#### Case 1: Low costs and high gains\n", "\n", "* If we have very low costs then we don't mind having more False Positives.\n", "* In cases like these, we don't mind low precision so that we can choose high recall values.\n", "* Let assume costs=USD 2 per failed customer and gains=USD 25 per succeeded customer.\n", "* So lets see what happens when we take a high recall value and low precision value.\n", "* Looking at the graph, lets select threshold of 0.05 corresponding to recall ~ 90%" ] }, { "cell_type": "code", "execution_count": 56, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 0, 2613],\n", " [ 0, 415]], dtype=int64)" ] }, "execution_count": 56, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Extracting indices of Probs. over 80 percentile or 8th Decilbe (Top 20% probabilities)\n", "rf_probs_top20percent_indices = np.where(rf_probs > 0.05)\n", "\n", "# Using those indices to get the true labels of those probs.\n", "rf_probs_top20percent_y_test = np.array(y_test)[rf_probs_top20percent_indices]\n", "\n", "# Separating top 20% probs in a separate array\n", "rf_probs_top20percent_probs = rf_probs[rf_probs_top20percent_indices]\n", "\n", "# creating all positive predictions for top 20% probabilities\n", "rf_probs_top20percent_preds = np.where(rf_probs_top20percent_probs > 0, 'term_deposit', 'no')\n", "\n", "confusion_matrix(rf_probs_top20percent_y_test, rf_probs_top20percent_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* Total costs = 5 x 999 = 5,226\n", "* Total gains = 415 x 25 = 10,375\n", "* Profit = 10,375 - 5,226 = 5149\n", "\n", "\n", "* Profit increased. So we can see that looking at our Avg. gains and costs, we can take a decision on precision and recall.\n", "\n", "#### Case 2: High costs and Low gains\n", "\n", "* In some cases we might have a high cost of taking an action.\n", "* In cases like these, we want to minimize the False Positives hence we want a high precision.\n", "* Let assume costs=USD 50 per failed customer and gains=USD 25 per succeeded customer.\n", "* So lets see what happens when we take a high precision value and low recall value.\n", "* Looking at the graph, lets select threshold of 0.475 corresponding to recall ~ 80%" ] }, { "cell_type": "code", "execution_count": 60, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 0, 26],\n", " [ 0, 59]], dtype=int64)" ] }, "execution_count": 60, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Extracting indices of Probs. over 80 percentile or 8th Decilbe (Top 20% probabilities)\n", "rf_probs_top20percent_indices = np.where(rf_probs > 0.475)\n", "\n", "# Using those indices to get the true labels of those probs.\n", "rf_probs_top20percent_y_test = np.array(y_test)[rf_probs_top20percent_indices]\n", "\n", "# Separating top 20% probs in a separate array\n", "rf_probs_top20percent_probs = rf_probs[rf_probs_top20percent_indices]\n", "\n", "# creating all positive predictions for top 20% probabilities\n", "rf_probs_top20percent_preds = np.where(rf_probs_top20percent_probs > 0, 'term_deposit', 'no')\n", "\n", "confusion_matrix(rf_probs_top20percent_y_test, rf_probs_top20percent_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* Profit = (59 x 25) - (26 x 50) = 1475 - 1300 = 175\n", "* This gives us very less profit.\n", "* Usually in real life this will never be the case because products and successes earn big sums.\n", "\n", "### Doing the same thing with gradient boosting result probabilities\n", "\n", "\n", "* Let us try GBM predicted probabilities.\n", "* For that we will first need a precision recall curve for GBM." ] }, { "cell_type": "code", "execution_count": 61, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0,0.5,'%')" ] }, "execution_count": 61, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "lbl_enc = LabelEncoder()\n", "y_test_encoded = lbl_enc.fit_transform(y_test)\n", "\n", "precision, recall, thresholds = precision_recall_curve(y_test_encoded, gbm_probs, pos_label=1)\n", "\n", "fig=plt.figure(figsize=(10, 8), dpi= 80, facecolor='w', edgecolor='k')\n", "thresholds = np.append(thresholds, 1)\n", "queue_rate = [] \n", "for threshold in thresholds:\n", " queue_rate.append((gbm_probs >= threshold).mean())\n", "plt.plot(thresholds, precision, color=sns.color_palette()[0])\n", "plt.plot(thresholds, recall, color=sns.color_palette()[1]) \n", "plt.plot(thresholds, queue_rate, color=sns.color_palette()[2])\n", "leg = plt.legend(('precision', 'recall', 'queue_rate'), frameon=True)\n", "leg.get_frame().set_edgecolor('k')\n", "plt.xlabel('threshold')\n", "plt.ylabel('%')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* Best precision recall combination is around 0.25" ] }, { "cell_type": "code", "execution_count": 66, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 0, 275],\n", " [ 0, 179]], dtype=int64)" ] }, "execution_count": 66, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Extracting indices of Probs. over 80 percentile or 8th Decilbe (Top 20% probabilities)\n", "gbm_probs_top20percent_indices = np.where(gbm_probs > 0.23)\n", "\n", "# Using those indices to get the true labels of those probs.\n", "gbm_probs_top20percent_y_test = np.array(y_test)[gbm_probs_top20percent_indices]\n", "\n", "# Separating top 20% probs in a separate array\n", "gbm_probs_top20percent_probs = gbm_probs[gbm_probs_top20percent_indices]\n", "\n", "# creating all positive predictions for top 20% probabilities\n", "gbm_probs_top20percent_preds = np.where(gbm_probs_top20percent_probs > 0, 'term_deposit', 'no')\n", "\n", "confusion_matrix(gbm_probs_top20percent_y_test, gbm_probs_top20percent_preds)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "* Profit = 3100.\n", "* This is better than the previous best deciles threshold we used for the GBM model" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Future Efforts\n", "\n", "**1. We validated our Profit results only on one test set, but it may happen that the results might change on a different test \n", "set. So can we create a system which will give us robust profit values? (Cross-validated profit values)**\n", "\n", "**2. Use LIME to Trust the predictive model which we will decide to implement in production.**\n", "\n", "**3. Use LIME to improve the model results if possible by looking at feature to response relationships locally.**" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "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": "3.6.5" } }, "nbformat": 4, "nbformat_minor": 2 }