question_2_part_a.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. Predicting Change in Sentiments\n",
    "\n",
    "The goal of this section is to predict the change in the sentiment given the current utterance and the sentiment of the previous utterance. The inputs of the prediction model are the utterance, encoded in a vector of real numbers, and the previous utterance sentiment, a real number between -1 and 1.\n",
    "\n",
    "To encode the current utterance, I used a 50-dimensional GloVe vector for each word in the message. I then computed the average of all the word vectors for the utterance to produce a 50-dimensional feature vector for the utterance. \n",
    "The input of the model has 51 features in total: 50 features for the utterance and 1 feature for the previous sentiment. The output of the model is a real number between -2 and +2 which is the difference between the current sentiment and the previous one. \n",
    "\n",
    "For my model I used linear regression, support vector regression, random forest and a recurrent neural network. \n",
    "\n",
    "## 1.1 Metric of Evaluation\n",
    "I used mean square error and r-squared to compare the performaces of different models.\n",
    "\n",
    "## 1.2 Models\n",
    "\n",
    "### 1.2.1 Linear Regression\n",
    "I trained a linear regression model with all the 51 features and recorded the mean square error and r-squared score for the test set. Looking at the coefficients of the features, I can see that the feature for the previous sentiment has a very large positive coefficient (~0.9) while the other features have a small coefficient (~0.0). It seems that the model is only looking at the previous sentiment's feature to predict the change in the sentiment while ignoring the incoming message's features. The r-squared score is low for the training error which means this model is too simple to learn from the input features.\n",
    "\n",
    "### 1.2.2 Linear Regression with Principal Component Analysis\n",
    "In the next experiment, I used principal component analysis to reduce the number of features from 51 to 34 while keeping 95 percent of the variance. However, I did not observe any increase or decrease in the performance of the linear regression model. \n",
    "\n",
    "### 1.2.3 Linear Regression with L1 Regularization \n",
    "In the next experiment, I used L1 regularization to reduce the number of features. However, I did not observe any increase or decrease in the performance of the linear regression model. \n",
    "\n",
    "### 1.2.4 Linear Regression with Downsampling \n",
    "The distribution of the training samples shows that class labels are unbalanced and more than 8K samples in the training set of ~67K samples have a class label of zero (no change in emotions). Therefore, I downsampled the training samples with the label of 0 from ~8K samples to ~1K samples. This downsampling of train dataset, however, did not change the performance of the linear regression model on the test set.\n",
    "\n",
    "### 1.2.5 Support Vector Regression\n",
    "I was not able to train a support vector regression (SVR) for this dataset even with a reduced number of features. The problem is the size of the dataset that is too large for SVR and SVR does not scale well with the size of data. I ran this model for 16 hours on my laptop with no success. \n",
    "\n",
    "### 1.2.6 Random Foreset Regression\n",
    "I trained a random forest with with all the 51 features. The model has a high R-squared value of 0.91 on the training set but a much smaller value of 0.58 on the test set. This difference means that the model is overfitting the training set. The overfitting problem could be resolved by tunning the hyper-parameters of the model. For example, adding more estimators or decreasing depth of trees could help with overfitting. \n",
    "\n",
    "### 1.2.7 Recurrent Neural Network (LSTM)\n",
    "The LSTM model performance is explained in [LSTM MODEL](question_2_part_b.ipynb)\n",
    "\n",
    "\n",
    "## 1.3 Results\n",
    "The results of the models on the training and test sets are shown in the following table. Since the mean square error of the LSTM model is the smallest, I chose it for my final model and I validated the LSTM model with the validation set where the mean square error is 0.017.\n",
    "\n",
    "model                    | MSE(test) | R^2(test) | MSE(train) | R^2(train)  \n",
    "-------------------------|-----------|-----------|------------|-----------\n",
    " Linear Reg              |   0.08    |   0.46    |   0.08     | 0.46\n",
    " Linear Reg (PCA)        |   0.08    |   0.46    |   0.08     | 0.46\n",
    " Linear Reg (L_1 Reg)    |   0.08    |   0.46    |   0.08     | 0.46\n",
    " Linear Reg Downsampling |   0.09    |   0.45    |   0.09     | 0.49\n",
    " Random Forest           |   0.07    |   0.58    |   0.01     | 0.91\n",
    " RNN (LSTM)              |   0.017   |   0.87    |   0.009    | 0.92"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "from utility import *\n",
    "\n",
    "from sklearn import metrics\n",
    "from sklearn import svm\n",
    "from sklearn import datasets, linear_model\n",
    "from sklearn.metrics import mean_squared_error, r2_score\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.utils import resample\n",
    "from sklearn.model_selection import GridSearchCV\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sentence_to_avg(sentence, word_to_vec_map):\n",
    "    '''\n",
    "    this function creates a vector of 50d for an input sentence\n",
    "    by average all the word vectors in the sentence\n",
    "    '''\n",
    "    sentence = re.sub(r'[^\\w\\s]', ' ', sentence.strip())  \n",
    "    words = [i.lower() for i in sentence.strip().split()]\n",
    "    avg = np.zeros((50,))\n",
    "    \n",
    "    if len(words) == 0: \n",
    "        print(sentence)\n",
    "        return None\n",
    "    \n",
    "    for w in words:\n",
    "        vec = word_to_vec_map[w] if w in word_to_vec_map else word_to_vec_map['unk'] \n",
    "        avg += vec\n",
    "    avg = avg / len(words)\n",
    "        \n",
    "    return avg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit_svm_model(x_train, y_train):\n",
    "    # train svm\n",
    "    \n",
    "    clf = svm.SVR(kernel='linear', C=1000)\n",
    "    clf.fit(x_train, y_train)\n",
    "    return clf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit_random_forest_model(x_train, y_train):\n",
    "    # train random forest\n",
    "    model = RandomForestRegressor(n_estimators=100, random_state=42) \n",
    "    model.fit(x_train, y_train)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit_regression_model(x_train, y_train):\n",
    "    # train linear regression\n",
    "    \n",
    "    model = linear_model.LinearRegression()\n",
    "    model.fit(x_train, y_train)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_features(samples_list):\n",
    "    '''\n",
    "    this function creates features for each conversation\n",
    "    '''\n",
    "    x, y = [], []\n",
    "    for sample in samples_list:\n",
    "        vec = sentence_to_avg(sample['utterance'], word_to_vec_map)\n",
    "        if vec is None: continue\n",
    "        \n",
    "        vec = vec.tolist()\n",
    "        vec.append(sample['current_emotion'])\n",
    "        \n",
    "        x.append(vec)\n",
    "        y.append(sample['current_emotion'] - sample['prev_emotion'])\n",
    "    \n",
    "    return x, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load glove.6B 50d\n",
    "word_to_index, index_to_word, word_to_vec_map = read_glove_vecs('glove.6B/glove.6B.50d.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "# loading train, test and validation sets\n",
    "train_conversations = load_conversations(category='train')\n",
    "find_sentiments(train_conversations)\n",
    "\n",
    "test_conversations = load_conversations(category='test')\n",
    "find_sentiments(test_conversations)\n",
    "\n",
    "validation_conversations = load_conversations(category='validation')\n",
    "find_sentiments(validation_conversations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   \n"
     ]
    }
   ],
   "source": [
    "train_samples = create_samples(train_conversations)\n",
    "x_train, y_train = create_features(train_samples)\n",
    "\n",
    "test_samples = create_samples(test_conversations)\n",
    "x_test, y_test = create_features(test_samples)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of training samples:  76051 76051 Number of features:  51\n"
     ]
    }
   ],
   "source": [
    "print('Number of training samples: ', len(x_train), len(y_train), 'Number of features: ', len(x_train[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a234d7320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# pca to reduce number of features with a variance = %95\n",
    "pca = PCA(n_components=0.95, svd_solver='full')\n",
    "pca.fit(x_train)\n",
    "new_x_train = pca.transform(x_train)   \n",
    "new_x_test = pca.transform(x_test)    \n",
    "plt.plot(np.cumsum(pca.explained_variance_ratio_))\n",
    "plt.xlabel('number of components')\n",
    "plt.ylabel('cumulative explained variance')\n",
    "plt.title(\"number of components vs. cumulative variance\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Coefficients: \n",
      " [-0.05518255 -0.09586653  0.02818487  0.01035777  0.0116708   0.04732483\n",
      "  0.00779986  0.06017013  0.02364771  0.00432414 -0.122796    0.09142667\n",
      "  0.01239393  0.00719745  0.02650578  0.03638743 -0.03397658  0.04212108\n",
      " -0.00732165  0.07830698 -0.0605946  -0.00650555 -0.032918    0.01033865\n",
      " -0.06733058 -0.00742075  0.01750965  0.02583285  0.02104933  0.03763537\n",
      "  0.01567542  0.03409339  0.02973543 -0.07539445 -0.00222797 -0.08600382\n",
      "  0.06424077 -0.00221389  0.02712701  0.07966255  0.03911491  0.01164392\n",
      "  0.01512918  0.01352975  0.03294507 -0.00647605  0.05839532  0.01001226\n",
      "  0.01446369 -0.06615325  0.9229181 ]\n",
      "Train Mean squared error: 0.08\n",
      "Train Variance score: 0.46 \n",
      "\n",
      "Test Mean squared error: 0.08\n",
      "Test Variance score: 0.46\n"
     ]
    }
   ],
   "source": [
    "# regression model\n",
    "rgModel = fit_regression_model(x_train, y_train)\n",
    "print('Coefficients: \\n', rgModel.coef_)    \n",
    "\n",
    "y_pred = rgModel.predict(x_train)\n",
    "print(\"Train mean squared error: %.2f\"% mean_squared_error(y_train, y_pred))\n",
    "print('Train variance score: %.2f \\n' % r2_score(y_train, y_pred))\n",
    "\n",
    "y_pred = rgModel.predict(x_test)\n",
    "print(\"Test mean squared error: %.2f\"% mean_squared_error(y_test, y_pred))\n",
    "print('Test variance score: %.2f' % r2_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Coefficients: \n",
      " [ 0.00000000e+00 -1.77529914e-02  7.31177687e-04  0.00000000e+00\n",
      " -0.00000000e+00  1.72254352e-04  2.62283367e-03  0.00000000e+00\n",
      "  0.00000000e+00 -0.00000000e+00  0.00000000e+00 -0.00000000e+00\n",
      " -1.93252703e-02 -0.00000000e+00 -0.00000000e+00  0.00000000e+00\n",
      " -0.00000000e+00 -0.00000000e+00 -0.00000000e+00  0.00000000e+00\n",
      " -0.00000000e+00 -0.00000000e+00  0.00000000e+00  0.00000000e+00\n",
      " -0.00000000e+00 -1.42922764e-02  4.88171484e-03  0.00000000e+00\n",
      "  0.00000000e+00  0.00000000e+00 -5.14059961e-04 -1.01634703e-02\n",
      "  0.00000000e+00 -0.00000000e+00  0.00000000e+00 -0.00000000e+00\n",
      "  0.00000000e+00 -1.94385506e-02  0.00000000e+00  1.12277663e-03\n",
      " -0.00000000e+00  0.00000000e+00  0.00000000e+00  4.48685309e-03\n",
      " -0.00000000e+00 -0.00000000e+00  0.00000000e+00  6.89713196e-03\n",
      "  0.00000000e+00  0.00000000e+00  8.85407958e-01]\n",
      "Train mean squared error: 0.08\n",
      "Train variance score: 0.46 \n",
      "\n",
      "Test mean squared error: 0.08\n",
      "Test variance score: 0.46\n"
     ]
    }
   ],
   "source": [
    "# linear regression with l_1 regularization\n",
    "regModel_Lasso = linear_model.Lasso(alpha=0.001)\n",
    "regModel_Lasso.fit(x_train, y_train)\n",
    "print('Coefficients: \\n', regModel_Lasso.coef_) \n",
    "\n",
    "y_pred = regModel_Lasso.predict(x_train)\n",
    "print(\"Train mean squared error: %.2f\"% mean_squared_error(y_train, y_pred))\n",
    "print('Train variance score: %.2f \\n' % r2_score(y_train, y_pred))\n",
    "\n",
    "y_pred_lasso = regModel_Lasso.predict(x_test)\n",
    "print(\"Test mean squared error: %.2f\"% mean_squared_error(y_test, y_pred_lasso))\n",
    "print('Test variance score: %.2f' % r2_score(y_test, y_pred_lasso))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Coefficients: \n",
      " [-0.01588763 -0.01961595  0.11603304  0.23784433 -0.21233813 -0.48977211\n",
      "  0.35331376 -0.09234219 -0.21640386  0.0998127  -0.30396224  0.2075275\n",
      "  0.07616187 -0.30565344 -0.00909076 -0.01723262 -0.03074026  0.18717432\n",
      " -0.07100299 -0.04446019 -0.03955126  0.08211716  0.06333477 -0.04804121\n",
      " -0.00297954 -0.02521183 -0.00679153  0.09885389 -0.08641928  0.01576869\n",
      "  0.01131894  0.03226107 -0.04156359  0.02640871]\n",
      "Train mean squared error: 0.08\n",
      "Train variance score: 0.46 \n",
      "\n",
      "Test mean squared error: 0.08\n",
      "Test variance score: 0.46\n"
     ]
    }
   ],
   "source": [
    "# regression model with a reduced number of features\n",
    "rgModel_PCA = fit_regression_model(new_x_train, y_train)\n",
    "print('Coefficients: \\n', rgModel_PCA.coef_)  \n",
    "\n",
    "y_pred = rgModel_PCA.predict(new_x_train)\n",
    "print(\"Train mean squared error: %.2f\"% mean_squared_error(y_train, y_pred))\n",
    "print('Train variance score: %.2f \\n' % r2_score(y_train, y_pred))\n",
    "\n",
    "y_pred = rgModel_PCA.predict(new_x_test)\n",
    "print(\"Test mean squared error: %.2f\"% mean_squared_error(y_test, y_pred))\n",
    "print('Test variance score: %.2f' % r2_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAGECAYAAAAMbeigAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAMgYwhkAgIwhnAEAyBjCGQCAjPn/zQC7L3UFQBYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a31794ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# distribution of change in emotion\n",
    "fig, ax = plt.subplots(figsize=(8, 6))\n",
    "sns.distplot(y_train, kde=True, color='green', hist_kws=dict(alpha=1))\n",
    "plt.xlabel('Change in Emotion: Emotion_<t> - Emotion_<t-1>')\n",
    "plt.title('Disribution of Outcome')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# downsample majority output: emotion=0\n",
    "data = {}\n",
    "data['input'] = x_train\n",
    "data['output'] = y_train\n",
    "data_df = pd.DataFrame(data)\n",
    "\n",
    "label_25 = data_df[data_df.output == 0.25]\n",
    "non_zero_label = data_df[data_df.output != 0]\n",
    "zero_label = data_df[data_df.output == 0]\n",
    "\n",
    "# Downsample majority label: emotion=0.0\n",
    "df_zeros_downsampled = resample(zero_label, \n",
    "                               replace=False,    # sample without replacement\n",
    "                               n_samples=label_25.count()[0],     # to match minority class\n",
    "                               random_state=123) # reproducible results\n",
    " \n",
    "# Combine minority class with downsampled majority class\n",
    "df_downsampled = pd.concat([df_zeros_downsampled, non_zero_label])\n",
    " "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downsampled:  input     58501\n",
      "output    58501\n",
      "dtype: int64\n",
      "Coefficients: \n",
      " [-0.08313461 -0.14098571 -0.01609979 -0.0042521   0.03715972  0.04615805\n",
      "  0.02822943  0.07511453  0.04154947  0.03692298 -0.17593751  0.08199837\n",
      " -0.07360039  0.00291884 -0.03698849  0.06862022 -0.09157225  0.07004675\n",
      " -0.02978599  0.13050075 -0.08057468 -0.07773773 -0.03190529  0.02704783\n",
      " -0.01083884  0.04302959  0.04119686  0.04029173  0.05226092  0.06125555\n",
      "  0.04468681  0.04546069  0.02474527 -0.11195073  0.01221713 -0.09152922\n",
      "  0.09014953 -0.03388063  0.04843823  0.01421851  0.04377269  0.07616073\n",
      "  0.07685781  0.05208835  0.0668411  -0.04680021  0.00273032  0.01504504\n",
      " -0.02617958 -0.08306238  0.99428566]\n",
      "Train mean squared error: 0.09\n",
      "Train variance score: 0.49 \n",
      "\n",
      "Test mean squared error: 0.09\n",
      "Test variance score: 0.45\n"
     ]
    }
   ],
   "source": [
    "# display new class counts\n",
    "print(\"Downsampled: \", df_downsampled.count())\n",
    "\n",
    "downsampled_y_train = df_downsampled['output']\n",
    "downsampled_x_train = df_downsampled['input']\n",
    "\n",
    "downsampled_x_train = downsampled_x_train.values.tolist()\n",
    "\n",
    "# regression model after downsampling\n",
    "rgModel = linear_model.LinearRegression()\n",
    "rgModel.fit(downsampled_x_train, downsampled_y_train)\n",
    "\n",
    "print('Coefficients: \\n', rgModel.coef_)   \n",
    "\n",
    "y_pred = rgModel.predict(downsampled_x_train)\n",
    "print(\"Train mean squared error: %.2f\"% mean_squared_error(downsampled_y_train, y_pred))\n",
    "print('Train variance score: %.2f \\n' % r2_score(downsampled_y_train, y_pred))\n",
    "\n",
    "y_pred = rgModel.predict(x_test)\n",
    "print(\"Test mean squared error: %.2f\"% mean_squared_error(y_test, y_pred))\n",
    "print('Test variance score: %.2f' % r2_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train mean squared error: 0.01\n",
      "Train variance score: 0.91 \n",
      "\n",
      "Test mean squared error: 0.07\n",
      "Test variance score: 0.58\n"
     ]
    }
   ],
   "source": [
    "# random forest with 51 features\n",
    "rfModel = RandomForestRegressor(max_depth=70, n_estimators=400)\n",
    "\n",
    "rfModel.fit(x_train, y_train)\n",
    "y_pred = rfModel.predict(x_train)   \n",
    "print(\"Train mean squared error: %.2f\"% mean_squared_error(y_train, y_pred))\n",
    "print('Train variance score: %.2f \\n' % r2_score(y_train, y_pred))\n",
    "\n",
    "y_pred = rfModel.predict(x_test)   \n",
    "print(\"Test mean squared error: %.2f\"% mean_squared_error(y_test, y_pred))\n",
    "print('Test variance score: %.2f' % r2_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 288 candidates, totalling 1440 fits\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100, total=  32.9s\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100, total=  32.8s\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100, total=  33.0s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100, total=  33.0s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=100, total=  31.5s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200, total= 1.1min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200, total= 1.1min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200, total= 1.1min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200, total= 1.1min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000, total= 6.4min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000, total= 6.4min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000, total= 6.5min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100, total=  44.5s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100, total=  40.4s\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000, total= 6.5min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100, total=  39.4s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100, total=  37.1s\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=100, total=  36.9s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=8, n_estimators=1000, total= 6.5min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300, total= 1.9min\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done  33 tasks      | elapsed: 18.3min\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000, total= 6.3min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000, total= 6.4min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000, total= 6.5min\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100, total=  39.5s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000, total= 6.6min\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100, total=  41.3s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100, total=  41.7s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100, total=  42.6s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=100, total=  42.9s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200, total= 1.4min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200, total= 1.3min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200, total= 1.3min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200, total= 1.3min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=200, total= 1.4min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300, total= 2.0min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300, total= 2.0min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=10, n_estimators=1000, total= 6.8min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300, total= 2.0min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000, total= 6.6min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000, total= 6.6min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100, total=  39.0s\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000, total= 6.7min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000, total= 6.7min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100, total=  38.3s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100, total=  38.5s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100, total=  37.6s\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=100, total=  38.0s\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200, total= 1.2min\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200, total= 1.2min\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=200, total= 1.2min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=3, min_samples_split=12, n_estimators=1000, total= 6.3min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300 \n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300, total= 1.8min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=1000 \n",
      "[CV]  bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=300, total= 1.9min\n",
      "[CV] bootstrap=True, max_depth=80, max_features=2, min_samples_leaf=4, min_samples_split=8, n_estimators=1000 \n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-111-a2ba8954151d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     15\u001b[0m                            cv=5, n_jobs=-1, verbose=2)\n\u001b[1;32m     16\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 17\u001b[0;31m \u001b[0mgrid_search\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     18\u001b[0m \u001b[0mbest_grid\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgrid_search\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbest_estimator_\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     19\u001b[0m \u001b[0mbest_grid\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/site-packages/sklearn/model_selection/_search.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y, groups, **fit_params)\u001b[0m\n\u001b[1;32m    637\u001b[0m                                   error_score=self.error_score)\n\u001b[1;32m    638\u001b[0m           for parameters, (train, test) in product(candidate_params,\n\u001b[0;32m--> 639\u001b[0;31m                                                    cv.split(X, y, groups)))\n\u001b[0m\u001b[1;32m    640\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    641\u001b[0m         \u001b[0;31m# if one choose to see train score, \"out\" will contain train score info\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/site-packages/sklearn/externals/joblib/parallel.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, iterable)\u001b[0m\n\u001b[1;32m    787\u001b[0m                 \u001b[0;31m# consumption.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    788\u001b[0m                 \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_iterating\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 789\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mretrieve\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    790\u001b[0m             \u001b[0;31m# Make sure that we get a last message telling us we are done\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    791\u001b[0m             \u001b[0melapsed_time\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_start_time\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/site-packages/sklearn/externals/joblib/parallel.py\u001b[0m in \u001b[0;36mretrieve\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    697\u001b[0m             \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    698\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0mgetattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_backend\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'supports_timeout'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 699\u001b[0;31m                     \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_output\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mextend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mjob\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    700\u001b[0m                 \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    701\u001b[0m                     \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_output\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mextend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mjob\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/multiprocessing/pool.py\u001b[0m in \u001b[0;36mget\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    636\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    637\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 638\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    639\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mready\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    640\u001b[0m             \u001b[0;32mraise\u001b[0m \u001b[0mTimeoutError\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/multiprocessing/pool.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    633\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    634\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 635\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_event\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    636\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    637\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtimeout\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    549\u001b[0m             \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_flag\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    550\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 551\u001b[0;31m                 \u001b[0msignaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_cond\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    552\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0msignaled\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    553\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/tensorflow/lib/python3.6/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m    293\u001b[0m         \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m    \u001b[0;31m# restore state no matter what (e.g., KeyboardInterrupt)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    294\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 295\u001b[0;31m                 \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    296\u001b[0m                 \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    297\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# hyper parameter search for random forest\n",
    "rfModel = RandomForestRegressor() \n",
    "\n",
    "param_grid = {\n",
    "    'bootstrap': [True],\n",
    "    'max_depth': [10, 80, 90, 100, 110],\n",
    "    'max_features': [2, 3],\n",
    "    'min_samples_leaf': [3, 4, 5],\n",
    "    'min_samples_split': [8, 10, 12],\n",
    "    'n_estimators': [100, 200, 300, 1000]\n",
    "}\n",
    "\n",
    "# Instantiate the grid search model\n",
    "grid_search = GridSearchCV(estimator=rfModel, param_grid=param_grid, \n",
    "                           cv=5, n_jobs=-1, verbose=2)\n",
    "\n",
    "grid_search.fit(x_train, y_train)\n",
    "best_grid = grid_search.best_estimator_\n",
    "best_grid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a23e4add8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# distribution of change in emotion after downsampling zeros\n",
    "fig, ax = plt.subplots(figsize=(8, 6))\n",
    "sns.distplot(downsampled_y_train, kde=True, color='green', hist_kws=dict(alpha=1))\n",
    "plt.xlabel('Change in Emotion: Emotion_<t> - Emotion_<t-1>')\n",
    "plt.title('Disribution of change in emotion after downsampling')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1a17e09fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# distribution of \n",
    "emotions_df = pd.DataFrame(train_samples)\n",
    "pd.plotting.scatter_matrix(emotions_df, figsize=(4, 4))\n",
    "# plt.title('distribution of emotions')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [default]",
   "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}