01_Linear_Regression_for_Practice_TensorFlow_.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Linear Regression for TensorFlow Practice"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 라이브러리 import"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 주어진 데이터 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x_data = [1, 2, 3, 4]\n",
    "y_data = [2, 3, 4, 5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. 그래프(Graph) 만들기 \n",
    "#### Placeholder 지정"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X = tf.placeholder(tf.float32, shape=[None], name=\"X\")\n",
    "Y = tf.placeholder(tf.float32, shape=[None], name=\"Y\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 최적화할 변수 지정(W, b)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "W = tf.Variable(tf.zeros([1]), name=\"weight\")\n",
    "b = tf.Variable(tf.zeros([1]), name=\"bias\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### H(x) =  X*W + b "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "hypothesis = X*W+b"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 손실함수 만들기(Cost Function)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "cost = tf.reduce_mean(tf.square(hypothesis - Y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learning_rate = 0.1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 경사 하강법(Gradient Descent Algorithm)을 통해 최소값 찾기\n",
    "- 1) 직접 작성"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# cost function 미분값(경사도)\n",
    "gradient_W = tf.reduce_mean(((W*X+b) - Y)* X)\n",
    "gradient_b = tf.reduce_mean(((W*X+b) - Y))\n",
    "# \n",
    "descent_W = W - learning_rate * gradient_W\n",
    "descent_b = b - learning_rate * gradient_b\n",
    "update_W = W.assign(descent_W)\n",
    "update_b = b.assign(descent_b)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 2) 경사 하강법 최적화 함수 사용(Gradient Descent Optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "optimizer = tf.train.GradientDescentOptimizer(learning_rate=learning_rate)\n",
    "train_step = optimizer.minimize(cost) # 경사 하강법의 최솟값 호출"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. 그래프(Graph) 실행\n",
    "#### 그래프 실행을 위한 세션 생성\n",
    "- 세션에서 모든 동작의 실행은 tf.run()으로 작동"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sess = tf.Session()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 변수 초기화"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sess.run(tf.global_variables_initializer())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 그래프 반복 실행"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Tensor Flow Gradient Descent Optimizer 함수 사용"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 13.5 [ 2.] [ 0.69999999]\n",
      "1 6.09 [ 0.6500001] [ 0.26000002]\n",
      "2 2.76135 [ 1.54500008] [ 0.58299994]\n",
      "3 1.26525 [ 0.93599987] [ 0.39389989]\n",
      "4 0.59203 [ 1.33505011] [ 0.54711998]\n",
      "5 0.288352 [ 1.0589149] [ 0.47017092]\n",
      "6 0.150677 [ 1.23545706] [ 0.54667926]\n",
      "7 0.087612 [ 1.10893178] [ 0.51961488]\n",
      "8 0.0581198 [ 1.18572664] [ 0.56122601]\n",
      "9 0.0437704 [ 1.12652373] [ 0.55611753]\n",
      "10 0.0362852 [ 1.15867937] [ 0.58163214]\n",
      "11 0.0319435 [ 1.12984431] [ 0.58596605]\n",
      "12 0.0290709 [ 1.14209473] [ 0.60385066]\n",
      "13 0.0269125 [ 1.12702727] [ 0.61203313]\n",
      "14 0.0251259 [ 1.1304698] [ 0.62611288]\n",
      "15 0.0235544 [ 1.12170863] [ 0.6356554]\n",
      "16 0.022125 [ 1.12131798] [ 0.64767003]\n",
      "17 0.020802 [ 1.11550593] [ 0.65747702]\n",
      "18 0.019567 [ 1.11350858] [ 0.66822863]\n",
      "19 0.0184092 [ 1.10913146] [ 0.67782861]\n",
      "20 0.0173218 [ 1.10651994] [ 0.68769711]\n",
      "21 0.0162994 [ 1.10289145] [ 0.69689775]\n",
      "22 0.0153377 [ 1.1001054] [ 0.70607245]\n",
      "23 0.0144329 [ 1.09691107] [ 0.71480525]\n",
      "24 0.0135816 [ 1.09414184] [ 0.72338867]\n",
      "25 0.0127805 [ 1.0912348] [ 0.73164004]\n",
      "26 0.0120266 [ 1.08856261] [ 0.73969465]\n",
      "27 0.0113173 [ 1.08587134] [ 0.74747443]\n",
      "28 0.0106498 [ 1.08332705] [ 0.75504386]\n",
      "29 0.0100216 [ 1.08081448] [ 0.76237154]\n",
      "30 0.00943053 [ 1.07840693] [ 0.76949]\n",
      "31 0.00887429 [ 1.07605147] [ 0.77638853]\n",
      "32 0.00835088 [ 1.07377994] [ 0.78308511]\n",
      "33 0.00785834 [ 1.07156742] [ 0.78957808]\n",
      "34 0.00739482 [ 1.06942725] [ 0.79587877]\n",
      "35 0.00695867 [ 1.06734693] [ 0.80198938]\n",
      "36 0.00654824 [ 1.06533182] [ 0.80791801]\n",
      "37 0.00616201 [ 1.06337512] [ 0.81366849]\n",
      "38 0.00579857 [ 1.06147814] [ 0.81924725]\n",
      "39 0.00545656 [ 1.05963731] [ 0.82465875]\n",
      "40 0.00513472 [ 1.05785191] [ 0.82990837]\n",
      "41 0.00483186 [ 1.0561198] [ 0.83500075]\n",
      "42 0.00454686 [ 1.05443978] [ 0.83994073]\n",
      "43 0.00427868 [ 1.05280972] [ 0.8447327]\n",
      "44 0.00402632 [ 1.05122876] [ 0.84938127]\n",
      "45 0.00378885 [ 1.04969501] [ 0.85389066]\n",
      "46 0.00356536 [ 1.04820728] [ 0.85826504]\n",
      "47 0.00335508 [ 1.0467639] [ 0.86250842]\n",
      "48 0.00315719 [ 1.0453639] [ 0.86662483]\n",
      "49 0.00297097 [ 1.04400563] [ 0.87061793]\n",
      "50 0.00279574 [ 1.04268825] [ 0.87449151]\n",
      "51 0.00263084 [ 1.04141009] [ 0.87824905]\n",
      "52 0.00247568 [ 1.04017043] [ 0.88189417]\n",
      "53 0.00232966 [ 1.03896773] [ 0.8854301]\n",
      "54 0.00219224 [ 1.03780103] [ 0.88886023]\n",
      "55 0.00206294 [ 1.03666937] [ 0.89218765]\n",
      "56 0.00194127 [ 1.03557146] [ 0.89541543]\n",
      "57 0.00182676 [ 1.03450656] [ 0.89854664]\n",
      "58 0.00171902 [ 1.03347337] [ 0.90158403]\n",
      "59 0.00161764 [ 1.0324713] [ 0.90453053]\n",
      "60 0.00152222 [ 1.03149903] [ 0.90738875]\n",
      "61 0.00143244 [ 1.03055608] [ 0.9101615]\n",
      "62 0.00134795 [ 1.02964127] [ 0.91285115]\n",
      "63 0.00126845 [ 1.02875376] [ 0.91546029]\n",
      "64 0.00119362 [ 1.02789307] [ 0.91799134]\n",
      "65 0.00112322 [ 1.02705789] [ 0.92044657]\n",
      "66 0.00105697 [ 1.02624774] [ 0.92282832]\n",
      "67 0.000994636 [ 1.02546203] [ 0.92513877]\n",
      "68 0.000935973 [ 1.02469957] [ 0.92737997]\n",
      "69 0.00088077 [ 1.02396023] [ 0.92955416]\n",
      "70 0.000828817 [ 1.02324283] [ 0.93166322]\n",
      "71 0.000779932 [ 1.02254701] [ 0.9337092]\n",
      "72 0.000733932 [ 1.02187192] [ 0.93569386]\n",
      "73 0.00069064 [ 1.02121711] [ 0.93761915]\n",
      "74 0.000649906 [ 1.02058184] [ 0.93948674]\n",
      "75 0.000611577 [ 1.01996565] [ 0.94129848]\n",
      "76 0.000575502 [ 1.01936793] [ 0.94305599]\n",
      "77 0.000541558 [ 1.01878798] [ 0.9447608]\n",
      "78 0.000509616 [ 1.01822555] [ 0.94641465]\n",
      "79 0.000479556 [ 1.01767993] [ 0.94801897]\n",
      "80 0.000451271 [ 1.01715052] [ 0.94957519]\n",
      "81 0.000424654 [ 1.01663709] [ 0.95108485]\n",
      "82 0.000399611 [ 1.01613903] [ 0.95254934]\n",
      "83 0.000376039 [ 1.01565576] [ 0.95396996]\n",
      "84 0.00035386 [ 1.01518714] [ 0.95534807]\n",
      "85 0.000332987 [ 1.01473248] [ 0.95668489]\n",
      "86 0.00031335 [ 1.01429129] [ 0.95798165]\n",
      "87 0.000294868 [ 1.01386356] [ 0.95923966]\n",
      "88 0.000277473 [ 1.01344848] [ 0.96045995]\n",
      "89 0.000261108 [ 1.01304579] [ 0.96164376]\n",
      "90 0.000245709 [ 1.01265526] [ 0.9627921]\n",
      "91 0.000231214 [ 1.01227641] [ 0.96390605]\n",
      "92 0.000217581 [ 1.01190877] [ 0.96498662]\n",
      "93 0.000204748 [ 1.01155233] [ 0.96603489]\n",
      "94 0.00019267 [ 1.01120639] [ 0.96705174]\n",
      "95 0.000181305 [ 1.01087093] [ 0.9680382]\n",
      "96 0.000170612 [ 1.01054549] [ 0.96899509]\n",
      "97 0.00016055 [ 1.01022971] [ 0.96992332]\n",
      "98 0.000151081 [ 1.00992346] [ 0.97082376]\n",
      "99 0.00014217 [ 1.00962639] [ 0.97169727]\n"
     ]
    }
   ],
   "source": [
    "for step in range(100):\n",
    "    _, cost_val = sess.run([train_step, cost], feed_dict={X: x_data, Y: y_data})\n",
    "    print(step, cost_val, sess.run(W), sess.run(b))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 13.5 [ 1.] [ 0.34999999]\n",
      "1 0.4225 [ 1.16250002] [ 0.41500002]\n",
      "2 0.0649593 [ 1.18687499] [ 0.43287501]\n",
      "3 0.0536404 [ 1.18849993] [ 0.44286877]\n",
      "4 0.0517909 [ 1.1864078] [ 0.4514569]\n",
      "5 0.050245 [ 1.18373775] [ 0.45970926]\n",
      "6 0.0487518 [ 1.18100715] [ 0.4678039]\n",
      "7 0.0473031 [ 1.17830086] [ 0.47577173]\n",
      "8 0.0458975 [ 1.17563224] [ 0.48361933]\n",
      "9 0.0445337 [ 1.1730032] [ 0.49134934]\n",
      "10 0.0432103 [ 1.17041349] [ 0.49896362]\n",
      "11 0.0419264 [ 1.16786242] [ 0.50646389]\n",
      "12 0.0406805 [ 1.1653496] [ 0.51385188]\n",
      "13 0.0394717 [ 1.16287446] [ 0.52112931]\n",
      "14 0.0382988 [ 1.16043627] [ 0.52829778]\n",
      "15 0.0371608 [ 1.15803456] [ 0.53535891]\n",
      "16 0.0360565 [ 1.15566897] [ 0.54231441]\n",
      "17 0.0349851 [ 1.15333867] [ 0.54916573]\n",
      "18 0.0339455 [ 1.15104318] [ 0.55591446]\n",
      "19 0.0329368 [ 1.14878213] [ 0.56256223]\n",
      "20 0.0319581 [ 1.14655495] [ 0.56911045]\n",
      "21 0.0310085 [ 1.14436114] [ 0.57556069]\n",
      "22 0.030087 [ 1.14220011] [ 0.58191431]\n",
      "23 0.029193 [ 1.14007139] [ 0.58817285]\n",
      "24 0.0283255 [ 1.13797462] [ 0.5943377]\n",
      "25 0.0274839 [ 1.1359092] [ 0.60041028]\n",
      "26 0.0266672 [ 1.13387465] [ 0.60639197]\n",
      "27 0.0258748 [ 1.13187063] [ 0.61228412]\n",
      "28 0.0251059 [ 1.12989664] [ 0.61808807]\n",
      "29 0.0243599 [ 1.1279521] [ 0.62380511]\n",
      "30 0.023636 [ 1.12603676] [ 0.62943655]\n",
      "31 0.0229337 [ 1.12415004] [ 0.63498372]\n",
      "32 0.0222522 [ 1.12229156] [ 0.64044785]\n",
      "33 0.021591 [ 1.12046099] [ 0.64583021]\n",
      "34 0.0209494 [ 1.11865771] [ 0.65113193]\n",
      "35 0.0203269 [ 1.11688137] [ 0.65635431]\n",
      "36 0.0197229 [ 1.11513174] [ 0.66149855]\n",
      "37 0.0191368 [ 1.11340833] [ 0.66656578]\n",
      "38 0.0185682 [ 1.11171067] [ 0.67155713]\n",
      "39 0.0180164 [ 1.1100384] [ 0.67647374]\n",
      "40 0.017481 [ 1.10839117] [ 0.68131673]\n",
      "41 0.0169616 [ 1.10676861] [ 0.68608725]\n",
      "42 0.0164576 [ 1.10517037] [ 0.69078636]\n",
      "43 0.0159686 [ 1.10359597] [ 0.69541514]\n",
      "44 0.0154941 [ 1.10204518] [ 0.69997466]\n",
      "45 0.0150336 [ 1.10051763] [ 0.70446587]\n",
      "46 0.0145869 [ 1.09901297] [ 0.7088899]\n",
      "47 0.0141535 [ 1.09753072] [ 0.71324766]\n",
      "48 0.0137329 [ 1.09607077] [ 0.7175402]\n",
      "49 0.0133248 [ 1.09463263] [ 0.7217685]\n",
      "50 0.0129289 [ 1.09321606] [ 0.72593349]\n",
      "51 0.0125447 [ 1.0918206] [ 0.73003614]\n",
      "52 0.0121719 [ 1.09044611] [ 0.73407739]\n",
      "53 0.0118103 [ 1.08909214] [ 0.73805809]\n",
      "54 0.0114593 [ 1.08775854] [ 0.74197924]\n",
      "55 0.0111188 [ 1.08644485] [ 0.74584168]\n",
      "56 0.0107884 [ 1.08515084] [ 0.74964631]\n",
      "57 0.0104678 [ 1.08387613] [ 0.75339395]\n",
      "58 0.0101568 [ 1.0826205] [ 0.7570855]\n",
      "59 0.00985496 [ 1.08138382] [ 0.76072186]\n",
      "60 0.00956212 [ 1.08016551] [ 0.76430374]\n",
      "61 0.00927798 [ 1.07896543] [ 0.76783198]\n",
      "62 0.00900228 [ 1.07778335] [ 0.77130741]\n",
      "63 0.00873479 [ 1.07661903] [ 0.7747308]\n",
      "64 0.00847523 [ 1.07547212] [ 0.77810299]\n",
      "65 0.0082234 [ 1.07434225] [ 0.78142464]\n",
      "66 0.00797904 [ 1.07322943] [ 0.78469664]\n",
      "67 0.00774194 [ 1.07213318] [ 0.78791964]\n",
      "68 0.00751188 [ 1.07105339] [ 0.79109436]\n",
      "69 0.00728867 [ 1.0699898] [ 0.79422158]\n",
      "70 0.00707209 [ 1.06894207] [ 0.79730201]\n",
      "71 0.00686196 [ 1.06790996] [ 0.8003363]\n",
      "72 0.00665804 [ 1.06689346] [ 0.80332518]\n",
      "73 0.0064602 [ 1.0658921] [ 0.80626929]\n",
      "74 0.00626823 [ 1.06490576] [ 0.80916935]\n",
      "75 0.00608197 [ 1.06393409] [ 0.81202596]\n",
      "76 0.00590125 [ 1.06297708] [ 0.81483984]\n",
      "77 0.0057259 [ 1.06203437] [ 0.81761158]\n",
      "78 0.00555576 [ 1.06110573] [ 0.82034183]\n",
      "79 0.00539067 [ 1.06019092] [ 0.82303119]\n",
      "80 0.00523049 [ 1.05928993] [ 0.82568032]\n",
      "81 0.00507506 [ 1.05840242] [ 0.82828981]\n",
      "82 0.00492426 [ 1.05752814] [ 0.8308602]\n",
      "83 0.00477793 [ 1.05666697] [ 0.83339214]\n",
      "84 0.00463596 [ 1.05581868] [ 0.83588618]\n",
      "85 0.0044982 [ 1.05498314] [ 0.83834291]\n",
      "86 0.00436453 [ 1.05416012] [ 0.84076285]\n",
      "87 0.00423485 [ 1.05334926] [ 0.8431465]\n",
      "88 0.004109 [ 1.05255067] [ 0.84549451]\n",
      "89 0.00398691 [ 1.05176401] [ 0.84780741]\n",
      "90 0.00386844 [ 1.05098915] [ 0.85008568]\n",
      "91 0.00375348 [ 1.05022585] [ 0.85232979]\n",
      "92 0.00364195 [ 1.049474] [ 0.85454035]\n",
      "93 0.00353373 [ 1.04873347] [ 0.85671782]\n",
      "94 0.00342873 [ 1.04800391] [ 0.85886264]\n",
      "95 0.00332685 [ 1.04728532] [ 0.86097538]\n",
      "96 0.00322799 [ 1.04657745] [ 0.86305654]\n",
      "97 0.00313207 [ 1.0458802] [ 0.86510652]\n",
      "98 0.003039 [ 1.04519343] [ 0.86712581]\n",
      "99 0.0029487 [ 1.04451692] [ 0.86911488]\n"
     ]
    }
   ],
   "source": [
    "for step in range(100):\n",
    "    _, _, cost_val = sess.run([update_W, update_b, cost], feed_dict={X: x_data, Y: y_data})\n",
    "    print(step, cost_val, sess.run(W), sess.run(b))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 세션 종료\n",
    "- 세션을 종료해야 그래프가 초기화가 됨"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sess.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}