{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Split BX\n",
    "\n",
    "This splits the BookCrossing data for a train-test sweep."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "library(readr)\n",
    "library(dplyr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "options(repr.matrix.max.rows = 20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Explicit Ratings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings = read_csv(\"build/bx-ratings.csv\", col_names = TRUE, col_types = 'iid') %>%\n",
    "    rename(user=userID, item=bookID)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dim(ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "explicit.test.users = ratings %>%\n",
    "    group_by(user) %>%\n",
    "    summarize(nratings=n()) %>%\n",
    "    filter(nratings >= 10) %>%\n",
    "    sample_n(5000)\n",
    "explicit.test.users\n",
    "dim(explicit.test.users)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ratings.group = ratings %>%\n",
    "    group_by(user) %>%\n",
    "    mutate(urid = sample(n())) %>%\n",
    "    ungroup()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test.ratings = explicit.test.users %>%\n",
    "    select(user) %>%\n",
    "    inner_join(ratings.group) %>%\n",
    "    filter(urid <= 5) %>%\n",
    "    select(user, item, rating)\n",
    "dim(test.ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train.ratings = explicit.test.users %>%\n",
    "    right_join(ratings.group) %>%\n",
    "    filter(urid > 5 | is.na(nratings)) %>%\n",
    "    select(user, item, rating)\n",
    "dim(train.ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "write_csv(test.ratings, \"build/bx-ratings-test.csv\")\n",
    "write_csv(train.ratings, \"build/bx-ratings-train.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Implicit Ratings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ratings = read_csv(\"build/bx-implicit.csv\", col_names = TRUE, col_types = 'iid') %>%\n",
    "    rename(user=userID, item=bookID)\n",
    "dim(ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "implicit.test.users = ratings %>%\n",
    "    group_by(user) %>%\n",
    "    summarize(nratings=n()) %>%\n",
    "    filter(nratings >= 10) %>%\n",
    "    sample_n(5000)\n",
    "implicit.test.users\n",
    "dim(implicit.test.users)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ratings.group = ratings %>%\n",
    "    group_by(user) %>%\n",
    "    mutate(urid = sample(n())) %>%\n",
    "    ungroup()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test.ratings = implicit.test.users %>%\n",
    "    select(user) %>%\n",
    "    inner_join(ratings.group) %>%\n",
    "    filter(urid <= 5) %>%\n",
    "    select(user, item, rating) %>%\n",
    "    mutate(rating=1)\n",
    "dim(test.ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train.ratings = implicit.test.users %>%\n",
    "    right_join(ratings.group) %>%\n",
    "    filter(urid > 5 | is.na(nratings)) %>%\n",
    "    select(user, item, rating)\n",
    "dim(train.ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "write_csv(test.ratings, \"build/bx-implicit-test.csv\")\n",
    "write_csv(train.ratings, \"build/bx-implicit-train.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "R",
   "language": "R",
   "name": "ir"
  },
  "language_info": {
   "codemirror_mode": "r",
   "file_extension": ".r",
   "mimetype": "text/x-r-source",
   "name": "R",
   "pygments_lexer": "r",
   "version": "3.4.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}