{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Loading data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "df = pd.read_parquet(\"fraud-all.parquet\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Cleaning data\n",
    "\n",
    "These data are mostly clean but we need to add a new field for transaction interarrival time."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.sort_values([\"user_id\", \"timestamp\"]).reset_index()\n",
    "del df['index']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "shifted = df.shift(1)[['user_id', 'timestamp']]\n",
    "\n",
    "df['prev_user_id'] = shifted['user_id']\n",
    "df['prev_timestamp'] = shifted['timestamp']\n",
    "df['interarrival'] = (df['timestamp'] - df['prev_timestamp']).where(df['user_id'] == df['prev_user_id'], np.NaN)\n",
    "\n",
    "del df['prev_user_id']\n",
    "del df['prev_timestamp']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "df[\"label\"] = df[\"label\"].astype(\"string\")\n",
    "df[\"trans_type\"] = df[\"trans_type\"].astype(\"string\")\n",
    "df[\"merchant_id\"] = df[\"merchant_id\"].astype(\"int32\")\n",
    "df[\"user_id\"] = df[\"user_id\"].astype(\"int32\")\n",
    "df[\"amount\"] = df[\"amount\"].astype(\"float32\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for this data set, brotli is ~60% the size of snappy (the default)\n",
    "# fastparquet can't currently handle strings (!)\n",
    "df.to_parquet(\"fraud-cleaned.parquet\", engine=\"pyarrow\", compression=\"brotli\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.sample(frac=0.05).to_parquet(\"fraud-cleaned-sample.parquet\", engine=\"pyarrow\", compression=\"brotli\")"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}