{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### METHOD - 1\n",
    "##### Train/Test Split\n",
    "\n",
    "# We train our model on the Train set and Evaluate on the Test set. \n",
    "# Typical fractions are 7:3 for train and test respectively.\n",
    "# Sklearn provides a method: train_test_split(X, Y, test_size=0.3, random_state=10) for this purpose\n",
    "# random_state is required to reproducing the results.\n",
    "# Very Fast"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### METHOD - 2\n",
    "##### K-Fold Cross Validation\n",
    "\n",
    "# We split our dataset into K folds (typical values are 3, 5, 10)\n",
    "# Algorithm is trained on K-1 folds, where 1 fold is held back and testing happens on that held back fold.\n",
    "# After running cross-validation you end up with k different performance scores that you can summarize\n",
    "# using a mean and a standard deviation.\n",
    "# Sklean provides KFold(n_splits=5, random_state=10) method for this purpose"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### METHOD - 3\n",
    "##### Leave One Out Cross Validation\n",
    "\n",
    "# We make 1 Fold of our dataset containing all the N datapoints.\n",
    "# We train our algorithm on N-1 points and predicts tha left out point.\n",
    "# N different  performance scores that you can summarize\n",
    "# Sklearn provides LeaveOneOut() method for this purpose\n",
    "# Computationaly intensive"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### METHOD - 4\n",
    "##### Repeated Random Train-Test Splits\n",
    "\n",
    "# Inspired by K-fold.\n",
    "# It's simply METHOD - 1 that is run N number of times with different random seed split of data.\n",
    "# Sklearn provides ShuffleSplit(n_splits=5, test_size=0.3, random_state=10) for this purpose"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}