Gradient descent-Multiple Linear Regression.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(100, 2)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data=np.loadtxt(\"data.csv\",delimiter=',')\n",
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "def step_gradient(points,learning_rate,m,c):\n",
    "    m_slope=0\n",
    "    c_slope=0\n",
    "    M=len(points)\n",
    "    for i in range(M):\n",
    "        x=points[i,0]\n",
    "        y=points[i,1]\n",
    "        m_slope+=(-2/M)*(y-m*x-c)*x\n",
    "        c_slope+=(-2/M)*(y-m*x-c)\n",
    "    new_m=m-learning_rate*m_slope\n",
    "    new_c=c-learning_rate*c_slope\n",
    "    return new_m,new_c"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gd(points,learning_rate,num_iterations):\n",
    "    m=0\n",
    "    c=0\n",
    "    for i in range(num_iterations):\n",
    "        m,c=step_gradient(points,learning_rate,m,c)\n",
    "        print(i,\"cost:\",cost(points,m,c))\n",
    "    return m,c    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cost(points,m,c):\n",
    "    total_cost=0\n",
    "    M=len(points)\n",
    "    for i in range(M):\n",
    "        x=points[i,0]\n",
    "        y=points[i,1]\n",
    "        total_cost+=(1/M)*((y-m*x-c)**2)\n",
    "    return total_cost\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run():\n",
    "    data=np.loadtxt(\"data.csv\",delimiter=',')\n",
    "    learning_rate=0.0001\n",
    "    num_iterations=100\n",
    "    m,c=gd(data,learning_rate,num_iterations)\n",
    "    print(m,c)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 cost: 1484.5865574086486\n",
      "1 cost: 457.8542575737672\n",
      "2 cost: 199.5099857255389\n",
      "3 cost: 134.50591058200533\n",
      "4 cost: 118.1496934223995\n",
      "5 cost: 114.0341490603815\n",
      "6 cost: 112.99857731713657\n",
      "7 cost: 112.73798187568467\n",
      "8 cost: 112.6723843590911\n",
      "9 cost: 112.65585181499745\n",
      "10 cost: 112.65166489759581\n",
      "11 cost: 112.6505843615011\n",
      "12 cost: 112.65028544701502\n",
      "13 cost: 112.65018320293967\n",
      "14 cost: 112.650130445072\n",
      "15 cost: 112.65009013922885\n",
      "16 cost: 112.6500529669463\n",
      "17 cost: 112.65001658353178\n",
      "18 cost: 112.64998039901865\n",
      "19 cost: 112.64994426496071\n",
      "20 cost: 112.64990814400622\n",
      "21 cost: 112.64987202675677\n",
      "22 cost: 112.64983591084761\n",
      "23 cost: 112.64979979568368\n",
      "24 cost: 112.64976368111523\n",
      "25 cost: 112.64972756710469\n",
      "26 cost: 112.64969145364236\n",
      "27 cost: 112.64965534072611\n",
      "28 cost: 112.64961922835512\n",
      "29 cost: 112.64958311652944\n",
      "30 cost: 112.64954700524868\n",
      "31 cost: 112.64951089451318\n",
      "32 cost: 112.64947478432279\n",
      "33 cost: 112.64943867467744\n",
      "34 cost: 112.64940256557728\n",
      "35 cost: 112.64936645702221\n",
      "36 cost: 112.64933034901203\n",
      "37 cost: 112.64929424154704\n",
      "38 cost: 112.64925813462712\n",
      "39 cost: 112.6492220282522\n",
      "40 cost: 112.64918592242235\n",
      "41 cost: 112.64914981713754\n",
      "42 cost: 112.64911371239779\n",
      "43 cost: 112.64907760820296\n",
      "44 cost: 112.64904150455324\n",
      "45 cost: 112.64900540144845\n",
      "46 cost: 112.64896929888867\n",
      "47 cost: 112.64893319687388\n",
      "48 cost: 112.6488970954041\n",
      "49 cost: 112.64886099447922\n",
      "50 cost: 112.64882489409929\n",
      "51 cost: 112.64878879426433\n",
      "52 cost: 112.64875269497436\n",
      "53 cost: 112.64871659622933\n",
      "54 cost: 112.64868049802914\n",
      "55 cost: 112.648644400374\n",
      "56 cost: 112.64860830326366\n",
      "57 cost: 112.64857220669828\n",
      "58 cost: 112.64853611067772\n",
      "59 cost: 112.64850001520212\n",
      "60 cost: 112.64846392027131\n",
      "61 cost: 112.64842782588545\n",
      "62 cost: 112.64839173204442\n",
      "63 cost: 112.6483556387483\n",
      "64 cost: 112.64831954599697\n",
      "65 cost: 112.64828345379043\n",
      "66 cost: 112.64824736212877\n",
      "67 cost: 112.64821127101193\n",
      "68 cost: 112.64817518043986\n",
      "69 cost: 112.64813909041264\n",
      "70 cost: 112.64810300093015\n",
      "71 cost: 112.64806691199259\n",
      "72 cost: 112.64803082359971\n",
      "73 cost: 112.64799473575155\n",
      "74 cost: 112.64795864844827\n",
      "75 cost: 112.64792256168963\n",
      "76 cost: 112.64788647547579\n",
      "77 cost: 112.64785038980668\n",
      "78 cost: 112.64781430468226\n",
      "79 cost: 112.64777822010265\n",
      "80 cost: 112.6477421360677\n",
      "81 cost: 112.64770605257743\n",
      "82 cost: 112.64766996963193\n",
      "83 cost: 112.64763388723107\n",
      "84 cost: 112.64759780537483\n",
      "85 cost: 112.64756172406335\n",
      "86 cost: 112.6475256432965\n",
      "87 cost: 112.64748956307432\n",
      "88 cost: 112.64745348339677\n",
      "89 cost: 112.64741740426388\n",
      "90 cost: 112.6473813256756\n",
      "91 cost: 112.64734524763193\n",
      "92 cost: 112.64730917013293\n",
      "93 cost: 112.6472730931785\n",
      "94 cost: 112.64723701676861\n",
      "95 cost: 112.64720094090339\n",
      "96 cost: 112.64716486558265\n",
      "97 cost: 112.64712879080662\n",
      "98 cost: 112.64709271657513\n",
      "99 cost: 112.64705664288809\n",
      "1.4788027175308358 0.035074970592341756\n"
     ]
    }
   ],
   "source": [
    "run()"
   ]
  }
 ],
 "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
}