random_state_SLEP.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.utils.validation import check_random_state\n",
    "from sklearn.base import clone"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Estimator():\n",
    "    def __init__(self, random_state=None):\n",
    "        self.random_state = _get_state(random_state)\n",
    "    \n",
    "    def fit(self):\n",
    "        rng = np.random.RandomState()\n",
    "        rng.set_state(self.random_state)\n",
    "        \n",
    "        print(rng.randint(0, 100))\n",
    "        \n",
    "    def get_params(self):\n",
    "        # return a RandomState instance whose state is\n",
    "        # self.random_state\n",
    "        random_state = np.random.RandomState()\n",
    "        random_state.set_state(self.random_state)\n",
    "        return {'random_state': random_state}\n",
    "    \n",
    "    def set_params(self, random_state=None):\n",
    "        # same as in __init__\n",
    "        self.random_state = _get_state(random_state)\n",
    "    \n",
    "def clone(est):  # just like base.clone\n",
    "    params = est.get_params()\n",
    "    return Estimator(**params)\n",
    "\n",
    "def _get_state(random_state):\n",
    "    rng = check_random_state(random_state)\n",
    "    \n",
    "    if random_state is None:\n",
    "        # rng is numpy's singleton. We don't want to use its state directly,\n",
    "        # because the singleton may not always be consumed between\n",
    "        # estimator instanciations.\n",
    "        BIG_INT = 1000000\n",
    "        rng = np.random.RandomState(rng.randint(0, BIG_INT))\n",
    "        \n",
    "    return rng.get_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "27\n",
      "27\n",
      "8\n",
      "8\n"
     ]
    }
   ],
   "source": [
    "# Multiple calls to fit/split on the same instances use the same rng\n",
    "a = Estimator(random_state=None)\n",
    "b = Estimator(random_state=None)\n",
    "\n",
    "a.fit()\n",
    "a.fit()\n",
    "\n",
    "b.fit()\n",
    "b.fit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "27\n",
      "27\n",
      "8\n",
      "8\n"
     ]
    }
   ],
   "source": [
    "# Clone has a natural behaviour\n",
    "# c and d have the same rng as a and b, respectively\n",
    "c = clone(a)\n",
    "d = clone(b)\n",
    "\n",
    "c.fit()\n",
    "c.fit()\n",
    "\n",
    "d.fit()\n",
    "d.fit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "27\n",
      "8\n"
     ]
    }
   ],
   "source": [
    "# Does not change a's rng\n",
    "a.set_params(random_state=a.get_params()['random_state'])\n",
    "a.fit()\n",
    "\n",
    "# Now a and b have the same rng state\n",
    "a.set_params(random_state=b.get_params()['random_state'])\n",
    "a.fit()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}