linear regression.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ライブラリ読み込み\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.datasets import load_boston\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ボストンデータ読み込み\n",
    "boston = load_boston()\n",
    "df = pd.DataFrame(boston.data, columns = boston.feature_names)\n",
    "df['target'] = boston.target\n",
    "\n",
    "#学習データ準備\n",
    "f_num = df.shape[1] - 1\n",
    "train_X = np.array(df.iloc[:, :f_num])\n",
    "train_Y = np.array(df.iloc[:, f_num: f_num + 1])\n",
    "n_samples = train_X.shape[0]\n",
    "\n",
    "#正規化\n",
    "ss = StandardScaler()\n",
    "ss.fit(train_X)\n",
    "train_X = ss.transform(train_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#データ\n",
    "with tf.name_scope('data'):\n",
    "    X = tf.placeholder(tf.float32, shape = [None, f_num ], name = \"X\")\n",
    "    Y = tf.placeholder(tf.float32, name = \"Y\")\n",
    "\n",
    "#パラメータ\n",
    "with tf.name_scope('parameter'):\n",
    "    W = tf.Variable(tf.zeros([f_num, 1]), name = \"weight\")\n",
    "    b = tf.Variable(tf.zeros([1]), name = \"bias\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#モデル\n",
    "with tf.name_scope('model'):\n",
    "    pred = tf.add(tf.matmul(X, W), b) #matmulは行列の積"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#損失関数\n",
    "with tf.name_scope('loss'):\n",
    "    # Mean squared error\n",
    "    loss = tf.reduce_mean(tf.square(pred - Y))\n",
    "    tf.summary.scalar('loss', loss)\n",
    "\n",
    "optimizer = tf.train.GradientDescentOptimizer(learning_rate = 0.01)\n",
    "train_step = optimizer.minimize(loss)\n",
    "\n",
    "#R2\n",
    "with tf.name_scope('r2'):\n",
    "    r2 = 1 - (tf.reduce_sum(tf.square(Y - pred)) / tf.reduce_sum(tf.square(Y - tf.reduce_mean(Y))))\n",
    "    tf.summary.scalar('r2', r2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#weight = 0\n",
    "#学習開始\n",
    "with tf.Session() as sess:\n",
    "    # ログの設定\n",
    "    summary = tf.summary.merge_all()\n",
    "    writer = tf.summary.FileWriter(\"boston_log\", sess.graph)\n",
    "    \n",
    "    sess.run(tf.global_variables_initializer())#変数初期化\n",
    "    \n",
    "    for i in range(1000):\n",
    "        sess.run(train_step, feed_dict={X: train_X, Y: train_Y})\n",
    "        if i != 0 and i % 200 == 0: # 200ステップごとに精度を出力\n",
    "            train_summary, train_loss, train_r2 = sess.run([summary, loss, r2], feed_dict={X: train_X, Y:train_Y})# コストと精度を出力\n",
    "            writer.add_summary(train_summary, i) #summaryの更新\n",
    "            \n",
    "            print(\"Step:\", '%04d' % (i), \"loss=\", \"{:.9f}\".format(train_loss), \"r2=\", \"{:.9f}\".format(train_r2), \"W=\", sess.run(W), \"b=\", sess.run(b))            \n",
    "            \n",
    "    training_cost, training_r2 = sess.run([loss,r2], feed_dict={X: train_X, Y: train_Y})\n",
    "    print(\"Training cost=\", training_cost, \"Training r2=\", training_r2, \"W=\", sess.run(W), \"b=\", sess.run(b), '\\n')\n",
    "    #weight = sess.run(W)\n",
    "    #pred_y = sess.run(pred, feed_dict={X: train_X})\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%bash\n",
    "tensorboard --logdir /Users/ogawashouhei/Google\\ ドライブ/Study/tf/boston_log --port 6017"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}