Solar Irradiance Prediction Based on Weather Patterns Using Bagging-Based Ensemble Learners with Principal Component Analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <p style='text-align: center;'> Solar Irradiance Prediction Based on Weather Patterns Using Bagging-Based Ensemble Learners with Principal Component Analysis </p>\n",
    "    \n",
    "<p style='text-align: justify;'>&emsp;Energy production of photovoltaic (PV) system depends on the amount of solar irradiance present on a certain location. Accurate prediction of solar irradiance ensures economic integration of PV system to grid and leads to optimal dispatching of available energy resources. Weather conditions has strong correlation with solar irradiance, and its erratic nature causes fluctuation to energy production. Therefore, it is difficult to achieve consistent optimal energy production and reliable prediction of solar irradiance </p>\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; In the study, a bagging-based ensemble learning system was used to predict solar irradiance based on weather patterns. Previous researches confirmed that ensemble learners produced unbiased and more accurate results compared to single learners. A pre-processed stacked long-short term memory model (stacked LSTM) was used as base learner in ensemble learning since it has good performance in handling time series sequences. A plot that compares the performance between single learner and ensemble learners was provided. </p>\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; From the plot, it shows that at some iteration, ensemble learners get consistent at providing more accurate predictions compared to single learners. Metrics used in the study include explained variance score, maximum residual error, mean absolute error, mean squared error, and regression score function.</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "from sklearn import metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "import tensorflow_docs as tfdocs\n",
    "import tensorflow_docs.plots\n",
    "import tensorflow_docs.modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense\n",
    "from tensorflow.keras.layers import Flatten\n",
    "from tensorflow.keras.layers import Dropout\n",
    "from tensorflow.keras.layers import LSTM\n",
    "from tensorflow.keras.layers import Bidirectional"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.utils import resample\n",
    "from matplotlib import pyplot\n",
    "from numpy import mean\n",
    "from numpy import std\n",
    "from numpy import array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import max_error\n",
    "from sklearn.metrics import r2_score\n",
    "from sklearn.metrics import mean_absolute_error\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import explained_variance_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def norm(x):\n",
    "    return (x - train_stats['mean']) / train_stats['std']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Long SHort Term Memory**\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; Long short-term memory is suitable for handling time series data because it is capable to store information at any arbitrary period. It is also faster to converge compared to recurrent neural network (RNN). LSTM consists of input gate it, output gate ot and forget gate ft which manage the information coming in and out of memory cell, and memory cell which is responsible for handling vanishing gradient problem. Each gate possesses activation function which receives the same input as input neurons. Rectified linear unit (ReLu) was used as an activation function because it is simpler to use compared to sigmoid function and produces better results. Stacked LSTM uses 2 LSTM layers with 64 neurons that serve as memory units.  </p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate_model(trainX, trainy, testX,testY):\n",
    "    \n",
    "    model = Sequential()\n",
    "    model.add(LSTM(64, activation='relu', return_sequences=True, input_shape=(4,1)))\n",
    "    model.add(LSTM(64, activation='relu'))\n",
    "    model.add(Dense(1))\n",
    "   \n",
    "    model.compile(loss='mse',\n",
    "                optimizer='adam',\n",
    "                metrics=['mae', 'mse'])\n",
    "    \n",
    "    early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', patience=10)\n",
    "    model.fit(trainX, trainy, epochs=1000, validation_split = 0.2, verbose=0, \n",
    "                    callbacks=[early_stop, tfdocs.modeling.EpochDots()])\n",
    "    \n",
    "    return model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "df= pd.read_csv(\"weather3.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Correlation Map**\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; The darkest color indicates strong inverse relationship, otherwise it possesses strong direct relationship between two variables. It is reported that solar irradiance correlates strongly with ambient light and ambient temperature, having a regression coefficient of 0.94 and 0.77, respectively. It implies that as the temperature goes warmer, the sunlight intensifies as well as the ambient light.</p>\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; Also, solar irradiance shows strong negative correlation with humidity, having a regression coefficient of -0.82. Humidity is associated with rainfall and high value of it is due to evaporation. Rainfall results to reduced sunlight intensity and ambient temperature. Lastly, solar irradiance has weak negative correlation with windspeed, having a regression coefficient of -0.19. Windspeed has weak direct relationship with humidity, sea level pressure and absolute pressure having regression coefficient values of 0.1836, 0.2161, and 0.2159, respectively. It means that there’s a chance of having reduced sunlight intensity if there is a strong windspeed due to high atmospheric pressure and humidity </p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2a146b01108>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,8))\n",
    "sns.heatmap(df.corr(),annot=True, linewidth=0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Split dataset into training and testing set\n",
    "\n",
    "train_dataset = df.sample(frac=0.8,random_state=0)\n",
    "test_dataset = df.drop(train_dataset.index)\n",
    "    \n",
    "train_stats = train_dataset.describe()\n",
    "train_stats.pop(\"Solar Irradiance\")\n",
    "train_stats = train_stats.transpose()\n",
    "    \n",
    "train_labels = train_dataset.pop('Solar Irradiance')\n",
    "test_labels = test_dataset.pop('Solar Irradiance')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "normed_train_data = norm(train_dataset)\n",
    "normed_test_data = norm(test_dataset)\n",
    "    \n",
    "train_final= normed_train_data.values\n",
    "test_final = normed_test_data.values\n",
    "target=train_labels.values\n",
    "test_target=test_labels.values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Principal Component Analysis**\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; In the study, principal component analysis was applied in order to reduce the size of input dataset without compromising its information. This also reduce the likelihood of overfitting the data, as well as minimizing the computational time requirement during model training </p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "pca = PCA(.95)\n",
    "pca.fit(train_final)\n",
    "    \n",
    "pca_train = pca.transform(train_final)\n",
    "pca_test = pca.transform(test_final)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = np.expand_dims(pca_train, axis=2)\n",
    "X_test = np.expand_dims(pca_test, axis=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Bagging Method**\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; A for-loop method was used in implementing the bagging method by setting a control number equal to the number of base learners. In the study, there are 20 base learners in an ensemble. Inside the iteration, the input dataset is resampled randomly and is later used for model training. The prediction made by the fitted model is saved on an array and the iteration will stop until it reached 20th iteration. The predictions made by ensemble learners in each iteration is averaged and is evaluated by metrics applicable in regression problems. A plot of performance comparison between ensemble learners and single learner is provided for further analysis </p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Epoch: 0, loss:204624.1250,  mae:335.3143,  mse:204624.1250,  val_loss:40370.2852,  val_mae:150.5346,  val_mse:40370.2852,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:8159.7734,  mae:50.6133,  mse:8159.7734,  val_loss:9712.5176,  val_mae:50.1953,  val_mse:9712.5176,  \n",
      "...............................\n",
      "Epoch: 0, loss:193808.3125,  mae:327.7590,  mse:193808.3125,  val_loss:40278.1484,  val_mae:148.7898,  val_mse:40278.1484,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:8128.5112,  mae:49.1543,  mse:8128.5112,  val_loss:8724.6094,  val_mae:48.0800,  val_mse:8724.6094,  \n",
      "......\n",
      "Epoch: 0, loss:190693.3750,  mae:322.2191,  mse:190693.3750,  val_loss:34934.9883,  val_mae:138.2930,  val_mse:34934.9883,  \n",
      "..........................................................................\n",
      "Epoch: 0, loss:195343.9844,  mae:326.8244,  mse:195343.9844,  val_loss:36226.2031,  val_mae:142.6552,  val_mse:36226.2031,  \n",
      "........................................................\n",
      "Epoch: 0, loss:194155.5312,  mae:326.0429,  mse:194155.5312,  val_loss:44682.0312,  val_mae:162.7998,  val_mse:44682.0312,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:8941.3623,  mae:53.3432,  mse:8941.3623,  val_loss:10014.4102,  val_mae:51.9187,  val_mse:10014.4102,  \n",
      "............\n",
      "Epoch: 0, loss:191136.1719,  mae:324.0061,  mse:191136.1719,  val_loss:35840.6875,  val_mae:136.6707,  val_mse:35840.6875,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:9334.3750,  mae:54.0374,  mse:9334.3750,  val_loss:9725.8203,  val_mae:51.2512,  val_mse:9725.8203,  \n",
      "................................................\n",
      "Epoch: 0, loss:205870.5156,  mae:337.3441,  mse:205870.5156,  val_loss:41237.8516,  val_mae:150.3844,  val_mse:41237.8516,  \n",
      "...................................................................................................\n",
      "Epoch: 0, loss:198515.7188,  mae:331.1545,  mse:198515.7188,  val_loss:43320.8047,  val_mae:158.9834,  val_mse:43320.8047,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:8188.3096,  mae:52.8315,  mse:8188.3096,  val_loss:8807.7549,  val_mae:46.8036,  val_mse:8807.7549,  \n",
      "......................................\n",
      "Epoch: 0, loss:209838.6094,  mae:343.7047,  mse:209838.6094,  val_loss:39738.0117,  val_mae:145.2663,  val_mse:39738.0117,  \n",
      "....................................................................................\n",
      "Epoch: 0, loss:212864.1094,  mae:344.7293,  mse:212864.1094,  val_loss:35409.5234,  val_mae:137.5256,  val_mse:35409.5234,  \n",
      ".................................................................................................\n",
      "Epoch: 0, loss:195880.7188,  mae:331.8269,  mse:195880.7188,  val_loss:45948.9961,  val_mae:166.2356,  val_mse:45948.9961,  \n",
      "..................................................................................................\n",
      "Epoch: 0, loss:199043.5781,  mae:333.1614,  mse:199043.5781,  val_loss:40849.3320,  val_mae:150.7089,  val_mse:40849.3320,  \n",
      "................................................................................................\n",
      "Epoch: 0, loss:191722.2344,  mae:323.6666,  mse:191722.2344,  val_loss:40154.0664,  val_mae:147.1510,  val_mse:40154.0664,  \n",
      ".................................................................................\n",
      "Epoch: 0, loss:203112.9688,  mae:333.6211,  mse:203112.9688,  val_loss:36344.7578,  val_mae:138.3769,  val_mse:36344.7578,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:9287.5723,  mae:54.2835,  mse:9287.5723,  val_loss:9674.4570,  val_mae:54.4441,  val_mse:9674.4570,  \n",
      "..........................................................................\n",
      "Epoch: 0, loss:203508.9844,  mae:334.3241,  mse:203508.9844,  val_loss:37534.4023,  val_mae:144.2695,  val_mse:37534.4023,  \n",
      "..................................................................\n",
      "Epoch: 0, loss:210774.1250,  mae:345.2138,  mse:210774.1250,  val_loss:37881.6172,  val_mae:141.2135,  val_mse:37881.6172,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:9091.5420,  mae:53.8713,  mse:9091.5420,  val_loss:9739.9990,  val_mae:54.3119,  val_mse:9739.9990,  \n",
      ".............................\n",
      "Epoch: 0, loss:207290.0156,  mae:335.7209,  mse:207290.0156,  val_loss:37384.3203,  val_mae:142.4419,  val_mse:37384.3203,  \n",
      "............................................................................\n",
      "Epoch: 0, loss:198686.0469,  mae:330.2875,  mse:198686.0469,  val_loss:36298.0430,  val_mae:143.7339,  val_mse:36298.0430,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:8619.4922,  mae:52.0937,  mse:8619.4922,  val_loss:8947.1641,  val_mae:50.0653,  val_mse:8947.1641,  \n",
      ".........................................................................\n",
      "Epoch: 0, loss:197646.4375,  mae:331.7383,  mse:197646.4375,  val_loss:44869.2969,  val_mae:162.0118,  val_mse:44869.2969,  \n",
      "...............................................................................................\n",
      "Epoch: 0, loss:220448.8594,  mae:357.5578,  mse:220448.8594,  val_loss:34286.1953,  val_mae:135.3437,  val_mse:34286.1953,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:8833.2383,  mae:52.6327,  mse:8833.2383,  val_loss:8514.5146,  val_mae:50.9946,  val_mse:8514.5146,  \n",
      "....................."
     ]
    }
   ],
   "source": [
    "n_splits = 20\n",
    "members = list()\n",
    "\n",
    "for _ in range(n_splits):\n",
    "    model = evaluate_model(X_train, target, X_test, test_target)\n",
    "    members.append(model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Results**\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; The following section shows the plot that shows comparison of the performance between single learners and ensemble learners using the metrics regression score, explained variance score, root mean squared error, mean absolute error, and maximum residual error. The plots for the said metrics are arranged from top to bottom. It can be observed from the plot the downward trend of ensemble learners at the 10th iteration. </p>\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; It implies that as iteration increases, the value of error starts to decrease. This can also imply the improvement of accuracy in prediction in comparison with the individual learners. It is noticeable the consistency of the performance of ensemble learners compared to single learner Using the mean absolute error and root mean squared error as metric, the ensemble learners became stable and consistent at 10th iteration. Meanwhile, by using the maximum residual error as y-axis, the ensemble learner starts to become consistent at around 6th iteration. </p>\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; There are some points of iterations where single learner outperforms ensemble learners, however, its performance is not consistent. At average, the explained variance score and regression score of ensemble learners average around 0.92, which indicates high estimation of true values found in input sequence. </p>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> 1: single=0.924, ensemble=0.924\n",
      "> 2: single=0.922, ensemble=0.929\n",
      "> 3: single=0.914, ensemble=0.927\n",
      "> 4: single=0.908, ensemble=0.924\n",
      "> 5: single=0.923, ensemble=0.925\n",
      "> 6: single=0.921, ensemble=0.926\n",
      "> 7: single=0.925, ensemble=0.927\n",
      "> 8: single=0.928, ensemble=0.929\n",
      "> 9: single=0.927, ensemble=0.930\n",
      "> 10: single=0.923, ensemble=0.930\n",
      "> 11: single=0.911, ensemble=0.929\n",
      "> 12: single=0.922, ensemble=0.929\n",
      "> 13: single=0.899, ensemble=0.929\n",
      "> 14: single=0.928, ensemble=0.929\n",
      "> 15: single=0.919, ensemble=0.929\n",
      "> 16: single=0.917, ensemble=0.929\n",
      "> 17: single=0.908, ensemble=0.929\n",
      "> 18: single=0.931, ensemble=0.929\n",
      "> 19: single=0.915, ensemble=0.929\n",
      "> 20: single=0.917, ensemble=0.929\n"
     ]
    }
   ],
   "source": [
    "single_scores, ensemble_scores = list(), list()\n",
    "for i in range(1, n_splits+1):\n",
    "    subset = members[:i]\n",
    "    yhats = [model.predict(X_test) for model in subset]\n",
    "    yhats = array(yhats)\n",
    "    averaged = np.average(yhats, axis=0)\n",
    "    ensemble = r2_score(test_target,averaged)\n",
    "    \n",
    "    *_, y1 = [members[i-1].predict(X_test)]\n",
    "    single =  r2_score(test_target,y1)\n",
    "    print('> %d: single=%.3f, ensemble=%.3f' % (i, single, ensemble))\n",
    "    single_scores.append(single)\n",
    "    ensemble_scores.append(ensemble)\n",
    "    \n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "r2_score 0.919 (0.008)\n",
      "r2_score 0.928 (0.002)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot score vs number of ensemble members\n",
    "print('r2_score %.3f (%.3f)' % (mean(single_scores), std(single_scores)))\n",
    "print('r2_score %.3f (%.3f)' % (mean(ensemble_scores), std(ensemble_scores)))\n",
    "x_axis = [i for i in range(1, n_splits+1)]\n",
    "pyplot.plot(x_axis, single_scores, marker='o', linestyle='None')\n",
    "pyplot.plot(x_axis, ensemble_scores, marker='o')\n",
    "plt.title(\"Coefficient of determination: Single Learner vs Ensemble Learners (Bagging)\")\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> 1: single=92.731, ensemble=92.731\n",
      "> 2: single=94.236, ensemble=89.402\n",
      "> 3: single=98.924, ensemble=91.077\n",
      "> 4: single=101.994, ensemble=92.750\n",
      "> 5: single=93.558, ensemble=91.885\n",
      "> 6: single=94.383, ensemble=91.592\n",
      "> 7: single=92.052, ensemble=90.777\n",
      "> 8: single=90.233, ensemble=89.571\n",
      "> 9: single=90.895, ensemble=89.153\n",
      "> 10: single=93.184, ensemble=89.042\n",
      "> 11: single=100.318, ensemble=89.635\n",
      "> 12: single=94.206, ensemble=89.594\n",
      "> 13: single=107.008, ensemble=89.874\n",
      "> 14: single=90.521, ensemble=89.685\n",
      "> 15: single=95.735, ensemble=89.906\n",
      "> 16: single=96.798, ensemble=89.782\n",
      "> 17: single=102.301, ensemble=89.951\n",
      "> 18: single=88.294, ensemble=89.484\n",
      "> 19: single=98.389, ensemble=89.493\n",
      "> 20: single=96.728, ensemble=89.518\n"
     ]
    }
   ],
   "source": [
    "single_mse, ensemble_mse = list(), list()\n",
    "for i in range(1, n_splits+1):\n",
    "    subset = members[:i]\n",
    "    yhats = [model.predict(X_test) for model in subset]\n",
    "    yhats = array(yhats)\n",
    "    averaged = np.average(yhats, axis=0)\n",
    "    ensemble = mean_squared_error(test_target,averaged,squared=False)\n",
    "    \n",
    "    *_, y1 = [members[i-1].predict(X_test)]\n",
    "    single =  mean_squared_error(test_target,y1,squared=False)\n",
    "    print('> %d: single=%.3f, ensemble=%.3f' % (i, single, ensemble))\n",
    "    single_mse.append(single)\n",
    "    ensemble_mse.append(ensemble)\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSE Single Learners 95.625 (4.615)\n",
      "RMSE Ensemble Learners 90.245 (1.120)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAc0AAAEICAYAAAA9YK8aAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3dd7wcVf3/8debJMANhHRKEiAIGFC6AUEUkQ6KBCw/EBQQBBS75gvoV8RGEflavxaa4BdEUDGgdKkWQC4EQigxdFJIbgihJUjK5/fHOQubze69e3Pv3ZJ9Px+PfezuzJmZs7Mz85lz5swZRQRmZmbWtdXqnQEzM7Nm4aBpZmZWJQdNMzOzKjlompmZVclB08zMrEoOmmZmZlVy0GwhkkLSZvXOR2+R9EtJ3+ilea1S68Yq6+y/lnSUpL/XOk/NRNJlkiY0QD4Ol3RjL83rSkn7VZO2y6Ap6SlJiyS9Iuk5SRdJWrsXMjk2b7z9O0lzWk7z+ZLhX8zDT+tpPrpL0rsl/VPSi5LmS/qHpB1rnY/eJuk2Sa/l/7nw+nMD5OsYSY9KelnSHEnXSBoEEBEnRMR36pw/H2SL5OPD6yXb0QP1zlct5H3o2Hrnoy9J2gbYFrgqfz9K0tKi//oJSZ+uRV4i4tKI2KeXZncm8L1qElZb0jwwItYGtgO2B05ZyYytjH8DR5YM+0QeXlOS1gH+AvwUGAaMBr4F/KcOeenXB7P9bESsXfQ6sMKyVzjR6ezkp9p5lEnzXuB04LCIGARsCVzRneWsCrq7bms9vzK+X7IdbdvHy2sZDbAtHA9cGsv3inNn4b8GPgx8X9L2vZbJGoiIfwHrSBrfVdpuVc9GxHPADaTgCYCkwZJ+I6lD0tOS/lvSanncavn705Lm5nSD86R35PcF+QxllwqLvQcYKOnteZ5vB9ry8DdI+oCk+yUtyCXBbYrGnSzp8VxaeVjSwUXjjpL0d0k/kPSCpCcl7V8hL2/N6+GyiFgaEYsi4saImJLn1S/PZ14+4zqxuDSdS+17FS37NEmXFH3/fS7NvyjpjsJvzuMukvQLSddKehV4n6Q18vKeyaWwX0pqK5pmoqTZkmZJ+mSF39QlSbtLmiHpJEnPAb8uNyyn/ZSkx3Ip/GpJo4rmE3mdTAemV7HoHUk75GSAiJgfERdHxMtF6+S7JXn8St7WZks6umjZwyX9WdJLku6R9N1KJcSu1ms31ttgSRfkvMzMy+yXx20q6RZJz+ft5VJJQ4qmfSqv2ynAq5L652FflTQlbyOXS1qzaJrO9oEV5leS119K+kHJsKskfTl/Pin/hpclTZO050qsj0Lt0pF53c6T9PWi8TtJas//0RxJ/1M0buf8mxZIekDS7kXjbsvr9p/KNST5/7606P8eW5KdA5T20XmSzlY+ZpXJ8xaSbsrb8zRJH+3u764i/0dLeiSv2yckHV80rrN9r9K2XnH7rTC/EZL+kvM2X9LfKq0PYH/g9kq/MyLuAx4hneAW8tPZca3T/VLSPnm9vyjp55JuVy7Nq6SWJ29bJ0iarnQs/19JyuP6STon/99PSvqsVqzpvA14f6XfVvwjO30BTwF75c9jgAeBHxeN/w2pqD4IGEsqAR6Tx30SeAx4C7A2cCXwf3ncWCCA/p0s+zTgEuBrwFl52PdJJd1LgNPysB2AucA7gX6kkulTwBp5/EeAUaSThP8HvApskMcdBSwGPpWn/TQwC1CZ/KwDPA9cTNp4hpaMPwF4FNiQVBK9tfg3Fq/L4t9X9P2TeT2uAfwIuL9o3EXAi8Cu+XesmdNcnZc1CPgzcEZOvx8wB9gKWAv4bc7LZhXW9W3AsRXG7Q4sAc7KeWurMGwPYF7+P9YglcjvKJpPADfl/LblYX8BTq6w3PcAi0il+V0L/2fJOvluSR6/DQwADgAWFv4j4Hf5NRB4G/As8PeSvG2WP1dcr2XyeFTxfErGTQJ+ldf/usC/gOPzuM2AvfN6Gkk6ifxRyX53P2lbaisa9i/StjyMdHA6ocp9YIX5leR1t7xOlL8Pzet+FDAujxtVtO9uWuE3v/GflBk3Nq/n8/L2si2plmbLPP5O4OP589rAzvnzaNJ+dwBp2987fx9ZtO0+BmwKDAYeJh2H9gL6k45Rvy75r2/N63CjnPbY0v8z/2/PAkfn+exA2r7f3p19qIr8vz/nXcB7SdvtDlXse5W29c6OC+XmdwbwyzyvAaT9rtzxb6287kZW2v5JJ7oLgLdWeVyruF8CI4CXgEPy+v8C6Vi9wn9V9L/+BRiS/9cOYL+iY/PDpBg2FPgrJfEH+DJwZaV49Ea6LhOkne0V4OW8kJuBIXlcP9JG/7ai9McDt+XPNwOfKRo3Lv/o/nQvaG4EPJP/0GdIO35x0PwF8J2SaacB760w3/uBg4pW/GNF4wbmfK1fYdotSQeGGXnjuxpYL4+7hXwQy9/3oRtBs2Q5Q/K0g4sORr8pGi9S8N+0aNguwJP584XAmUXj3krXQXMhaYMvvL5TtKO9DqxZlL7csAtIVXOF72vn/3ts0Ua9R1fbXEm+9ift9Avydvg/QL+idVIcNBex/E4wF9iZtJ0uBsYVjfsuZYJmV+u1TP6OokzQBNYj7RttRcMOA26tMJ8JwOSS/e6TZfbFI4q+fx/4ZTX7QLn5laQVad/aLX//FHBL/rxZXpd7AQO6+L8uAl4r2Y4uzuPG5vU8pij9v4BD8+c7SCdII0rmeRL5ZLto2A3AkUXb7teLxp0DXFf0/UCWP1AH+WCav38GuLn0/ySdYP+tZLm/Ar7ZyT5ULmh2mv8y6ScBX+hi36u0rXd1XCg3v2+TCj5ljw1F6UbndVc87VGk42Bh/wzSyfIKQTenf+O4Rhf7Jeky3J0l2+izdB403130/QryCTnp2Hx80bi9WDFovrHNd/aqtnp2QqRrSrsDW5DOAMjvqwNPF6V9Oq9cSGeppeP6kw4oVYuIZ0hnkqcD0yPi2ZIkGwNfydULCyQtIAXWUQCSPlFUbbWAVPoaUTT9c0XLWpg/lm3sFBGPRMRRETEmz2cU6eyp8HuL8/Z06fSV5OqDM5WqkV8iHeQoyWfxvEeSAvy9Rb/r+jx8ZfPy+YgYUvQqbpnaERGvlaQvHbbc/x0Rr5DOqEcXpSn97zoVEddFurY6DDiItKNUamzxfEQsKfq+kPQ/jiRtd8XLrpSPrtZrtTYmneTNLprPr0glTiStK+l3SlWeL5FOAkeUzKNcHp8r+lz4fYXlVdwHOpkfAJGOGr8jBXaAjwGX5nGPAV8kneTNzfkeVW4+2Q9KtqMjq/wNx5BO7h7NVXUfKPptHyn5be8GNiiaz5yiz4vKfC/dn0v3jXK/Z2PgnSXLPRxYv9yP7kSn+Ze0v6S7ctXoAlLJsXhbKLfvdbatd7X9ls7vbNLx9cZcPXxyhd+xIL8PKhl+V/6f1yatm7eTjtVdHde62i+XO4blbXRGhbwVVNq2So+H5faFQbz5Gyvq7jXN20lnkoVrH/NIZwobFyXbCJiZP88qM24JaYOO7iybVMXylfxe6lngeyU76sCIuEzSxqTqoM8CwyNiCDCVdNbSIxHxKGl9bJUHzSYdqAo2KpnkVdIGXVC8832MFBT2Ip2Fjc3Di/NZvM7mkQ4Gby/6zYPzhltNXrqr3P9VOmy5/1vSWsBw3tweKs2n64VHLIuIm0lnjFt1lb5EB2m7G1M0bMMKabtar9V6llTSHFE0n3UionA95wzSutgmItYBjmDFbbI766riPtCN+V0GfDjvM+8E/vjGhBG/jYh3k/7fIFXv9aqImB4Rh5FOLM4C/pC3oWdJJbXi37ZWRJzZg8WV7huzyqR5Fri9ZLlrR0R3W4dWzL+kNUjr+QekGqshwLVU3u+7Us32u9z8IuLliPhKRLyFVCr/sspcs46IV4HHyW07yomIOfn3FBoRdnZc62q/nF08Ll+fLE7bHcvNi/L7/5ZAly29V+Y+zR8Be0vaLiKWkorA35M0KO9sXyadNUPaCb8kaROl21ROBy7PZ0gdwDLS9c5qXE6q7izXevI84ARJ71SylqT3K92aUKiH74B00Z3uH3TJ026RL76Pyd83JJ2Z35WTXAF8XtIYSUOB0jO2+4FDJQ1QaqX14aJxg0gH2edJgfX0zvISEcvy7/6hpELpZbSkfYvycpSkt0kaCHxzZX5zN/0WOFrSdvlgcDpwd0Q8tTIzk3SQpEMlDc3/606kaz53dTVtsbydXgmcJmmgpC1IVT/l0na1XitkVWsWvyJiNnAjcI6kdZQaxW2q1CIY0v/9Cqkh3GhgYnd+Uxmd7QNVidTgqgM4H7ghIhbkHzdO0h75P32NdFBe2sP8rkDSEZJG5v+gcMa/lHQ8OVDSvrnksqZSg5aVPYACTMzb1Yaka2WXl0nzF+Ctkj6e99kBknaUtGWZtAX9S7aFAV3kf3XStb4OYIlSI8SVvo1iZbZfpQZkm+Wg9BJpnVf6f68l7YOV5jUcOBh4KA+qeFyrYr+8Btha0gSlBjsn0v1SfsEVwBfyuhhCqjIv9V7guq5m1O2gGREdpNJeoeruc6QS1BPA30kHzgvzuAuB/yNdq3iStMN9Ls9nIem+mH/kaoSdu1juooj4a0QsKjOunVQf/TPgBVJVw1F53MOkaxx3kkq4WwP/6O7vzl4mnYHfrdSC9S5SqfUrefx5pGsVDwD3kTaIYt8gXfB/gXTt5rdF435DqiaaSbpgXU1gOIn0W+/KVR9/JV03JiKuI53g3JLT3FLF/H6m5e+vu7eKad6QS4LfIJ1pzib91kM7m0bSdZK+VmH0C6T/dTppZ74EODsiLu1OvrLPks50nyNtk5dR+Vahiuu1gneRAskbr7yTf4J0UHw4/5Y/8GaV4rdIDUteJB0cSreVbulsH+imy0ilguJtcw3SfWzzSOtvXVLjvEr+q2Q7mlflsvcDHpL0CvBj0rXO1/LlmIPyMjtIJbeJ9KxzlquAe0knsteQrscvJ1Ir7X1I2/As0m8vNKCp5Bcsvy38urP852V8nnRQf4FUMru6B78Lur/9bp7TvEI6Tv48Im6rkPZc4PAcYAt2KfzXpMZpHeTjPF0f1yrulxExj9SI8/ukoPs2oJ2Vu8XvPNJJ7BRgMin4LyGfHCjda/9qpFtPOlVoKWd9QKmZ+5OkxhNLOk9ttSTpLFJjr9LrbWbWCUm/Ba6IiEl9MO+K+6XSbTAzgMMj4tYeLmd/UiO6jfP3PwIXRMS1XU3b1zc5mzWEXPWzOumWqR1JjU5W6d5bzPpCRHyst+bV1X6Zq5XvJpXaJ5KuhXbr8kyeTxvwPlJpcz3S5ao/FcZHxIeqnZeDprWKQaSqn1Gk5vnnkLsCM7O66Wq/3IV0qaBwmWNCuUt0VRDpksjlpAB8DXDqymTY1bNmZmZV8lNOzMzMqtTS1bMjRoyIsWPH1jsbZmZN5d57750XEd3t8GOV0NJBc+zYsbS3t9c7G2ZmTUVS1b2drWrqUj0r6UKl3vmnFg0bpvQ0gen5fWgePlGpC7z7JU1VenbbsDLzvEip9/pC2u1K05iZmfVEva5pXkS6kbnYyaROkzcndfR+MkBEnB0R20XEdqSnm9weEfMrzHdiIW1E3N9HeTczsxZVl6AZEXcApYHvINIjt8jvE8pMehipebKZmVnNNVLr2fVyf53k93WLRyr1n7ofRZ1Il/E9pQf0/jD3k7kCSccpPei2vaOjo7fybmZmLaCRgmZXDgT+0UnV7Cmkx5btSHqMVLkOeYmIcyNifESMHzmyJRt/mZnZSmqk1rNzJG0QEbMlbUDqHaLYoXRSNVsopQL/kfRr4Kt9lE+zljZp8kzOvmEasxYsYtSQNibuO44J24/uekKzVUAjlTSvBgqd9B5JUVdKkgaTHttSsduzHGgLz1ybQHr6iJn1okmTZ3LKlQ8yc8EiApi5YBGnXPkgkybP7HJas1VBvW45uYz0CJpxkmZIOob06KG9JU0H9s7fCw4GbswPQS2ez7V68wnyl0p6kNTx7wjgu339O8xazdk3TGPR4uUftbho8VLOvmFanXJkVlt1qZ7NT2cvZ4Wnhef0F5FuUykdfkDR5z16I29mVtmsBeX7yq403GxV00jVs2bW4EYNaevWcLNVjYOmmVVt4r7jaBvQb7lhbQP6MXHfcXXKkVltNVLrWTNrcIVWsm49a63KQdPMumXC9qMdJK1luXrWzMysSg6aZmZmVXLQNDMzq5KDppmZWZUcNM3MzKrkoGlmZlYlB00zM7MqOWiamZlVyUHTzMysSg6aZmZmVXLQNDMzq5KDppmZWZUcNM3MzKpUl6Ap6UJJcyVNLRo2TNJNkqbn96F5+O6SXpR0f36dWmGem0i6O09/uaTVa/V7zMysNdSrpHkRsF/JsJOBmyNic+Dm/L3gbxGxXX59u8I8zwJ+mKd/ATiml/NsZmYtri5BMyLuAOaXDD4IuDh/vhiYUO38JAnYA/jDykxvZmZWjUa6prleRMwGyO/rFo3bRdIDkq6T9PYy0w4HFkTEkvx9BlD2KbmSjpPULqm9o6OjN/NvZmaruEYKmpXcB2wcEdsCPwUmlUmjMsOi3Mwi4tyIGB8R40eOHNmL2TQzs1VdIwXNOZI2AMjvcwEi4qWIeCV/vhYYIGlEybTzgCGS+ufvY4BZtcm2mZm1ikYKmlcDR+bPRwJXAUhaP1+zRNJOpDw/XzxhRARwK/Dh0unNzMx6S71uObkMuBMYJ2mGpGOAM4G9JU0H9s7fIQXCqZIeAH4CHJqDJJKulTQqpzsJ+LKkx0jXOC+o3S8yM7NWoBx/WtL48eOjvb293tkwM2sqku6NiPH1zkc9NFL1rJmZWUNz0DQzM6uSg6aZmVmVHDTNzMyq5KBpZmZWJQdNMzOzKjlompmZVclB08zMrEoOmmZmZlVy0DQzM6uSg6aZmVmVHDTNzMyq5KBpZmZWJQdNMzOzKjlompmZVclB08zMrEoOmmZmZlVy0DQzM6tSXYKmpAslzZU0tWjYMEk3SZqe34fm4YdLmpJf/5S0bYV5XiTpSUn359d2tfo9ZmbWGupV0rwI2K9k2MnAzRGxOXBz/g7wJPDeiNgG+A5wbifznRgR2+XX/b2cZzMza3H967HQiLhD0tiSwQcBu+fPFwO3ASdFxD+L0twFjOnj7JmZVTRp8kzOvmEasxYsYtSQNibuO44J24+ud7asRhrpmuZ6ETEbIL+vWybNMcB1nczje7ka94eS1iiXQNJxktoltXd0dPQ812bWMiZNnskpVz7IzAWLCGDmgkWccuWDTJo8s95ZsxpppKDZKUnvIwXNkyokOQXYAtgRGFYpXUScGxHjI2L8yJEj+ySvZrZqOvuGaSxavHS5YYsWL+XsG6bVKUdWa3Wpnq1gjqQNImK2pA2AuYURkrYBzgf2j4jny01cKKUC/5H0a+CrfZ7jJuSqJbOVN2vBom4Nt1VPI5U0rwaOzJ+PBK4CkLQRcCXw8Yj4d6WJc6BFkoAJwNRKaVuVq5bMembUkLZuDbdVT71uObkMuBMYJ2mGpGOAM4G9JU0H9s7fAU4FhgM/z7eStBfN51pJo/LXSyU9CDwIjAC+W6Of0zRctWTWMxP3HUfbgH7LDWsb0I+J+46rU46s1urVevawCqP2LJP2WODYCvM5oOjzHr2Tu1WXq5bMeqZwKcOXOFpXI13TtD42akgbM8sESFctmVVvwvajHSRbWCNd07Q+5qolM7OecUmzhbhqycysZxw0W4yrlszMVp6rZ83MzKrkoGlmZlYlB00zM7MqOWiamZlVyUHTzMysSg6aZmZmVXLQNDMzq5KDppmZWZUcNM3MzKrkoGlmZlYld6Nn1mQmTZ7p/oPN6sRB06yJTJo8k1OufPCNh4nPXLCIU658EMCB06wGXD1r1kTOvmHaGwGzYNHipZx9w7Q65cistdQlaEq6UNJcSVOLhg2TdJOk6fl9aB4uST+R9JikKZJ2qDDPd0h6MKf7iSTV6veY1cqsMg8R72y4mfWuepU0LwL2Kxl2MnBzRGwO3Jy/A+wPbJ5fxwG/qDDPX+TxhbSl8zdreqOGtHVruJn1rroEzYi4A5hfMvgg4OL8+WJgQtHw30RyFzBE0gbFE+bv60TEnRERwG+KpjdbZUzcdxxtA/otN6xtQD8m7juuTjkyay2NdE1zvYiYDZDf183DRwPPFqWbkYcVG52Hd5YGAEnHSWqX1N7R0dErGTerlQnbj+aMQ7Zm9JA2BIwe0sYZh2ztRkBmNdIMrWfLXZuMlUiTBkacC5wLMH78+LJpzBrZhO1HO0ia1UkjlTTnFKpd8/vcPHwGsGFRujHArJJpZ+ThnaUxMzPrkUYKmlcDR+bPRwJXFQ3/RG5FuzPwYqEatyB/f1nSzrnV7CeKpjczM+sV9brl5DLgTmCcpBmSjgHOBPaWNB3YO38HuBZ4AngMOA/4TNF87i+a7aeB83O6x4Hr+vp3mJlZa6nLNc2IOKzCqD3LpA3gxArz2a7oczuwVa9k0MzMrIxGqp41MzNraA6aZmZmVXLQNDMzq1Iz3KdpRfxYKDOz+nHQbCJ+LJSZWX25eraJ+LFQZmb15aDZRPxYKDOz+nLQbCJ+LJSZWX05aDYRPxbKzCZNnsmuZ97CJidfw65n3sKkyTPrnaWW4oZATaTQ2MetZ81akxsD1p+DZpPxY6HMWldnjQF9XKgNV8+amTUJNwasPwdNM7Mm4caA9eegaWbWJNwYsP58TdPMrEm4MWD9OWiamTURNwasL1fPmpmZVclB08zMrEoNVT0r6QvApwAB50XEjyRdDhSucg8BFkTEdmWmfQp4GVgKLImI8bXJtZmZtYqGCZqStiIFzJ2A14HrJV0TEf+vKM05wIudzOZ9ETGvb3NqZmatqpGqZ7cE7oqIhRGxBLgdOLgwUpKAjwKX1Sl/ZmbW4hopaE4FdpM0XNJA4ABgw6Lx7wHmRMT0CtMHcKOkeyUdV2khko6T1C6pvaOjo9cyb2Zmq76GqZ6NiEcknQXcBLwCPAAsKUpyGJ2XMneNiFmS1gVukvRoRNxRZjnnAucCjB8/PnrtB5iZ2SqvkUqaRMQFEbFDROwGzAemA0jqDxwCXN7JtLPy+1zgT6Rro2ZmZr2moYJmLiUiaSNSkCyULPcCHo2IGRWmW0vSoMJnYB9Sda+ZmVmvaZjq2eyPkoYDi4ETI+KFPPxQSqpmJY0Czo+IA4D1gD+ltkL0B34bEdfXLttm1iwmTZ7pbuhspTVU0IyI91QYflSZYbNIjYWIiCeAbfs0c2bW9PwQZ5809FRDVc+amfWlzh7i3AoKJw0zFywiePOkYdLkmfXOWtNw0DSzltHqD3Fu9ZOG3uCgaWYto9Uf4tzqJw29wUHTzFpGqz/EudVPGnqDg6aZtYwJ24/mjEO2ZvSQNgSMHtLGGYds3TINYVr9pKE3NFTrWTOzvtbKD3Eu/G63nl15DppmZi2klU8aeoOrZ83MzKrkoGlmZlYlV8+aWVNxjzZWTw6aZtY03A2e1ZurZ82sabhHG6s3B00zaxru0cbqzUHTzJqGe7SxenPQNLOm4R5trN7cEMjMmoZ7tLF6c9A0s6biHm2snhquelbSFyRNlfSQpC/mYadJminp/vw6oMK0+0maJukxSSfXNudmZraqa6iSpqStgE8BOwGvA9dLuiaP/mFE/KCTafsB/wvsDcwA7pF0dUQ83MfZNjOzFtFoJc0tgbsiYmFELAFuBw6uctqdgMci4omIeB34HXBQH+XTzMxaUKMFzanAbpKGSxoIHABsmMd9VtIUSRdKGlpm2tHAs0XfZ+Rhy5F0nKR2Se0dHR29nX8zM1uFNVTQjIhHgLOAm4DrgQeAJcAvgE2B7YDZwDllJle5WZZZxrkRMT4ixo8cObK3sm5mZi2goYImQERcEBE7RMRuwHxgekTMiYilEbEMOI9UFVtqBm+WSgHGALP6PsdmZtYqGi5oSlo3v28EHAJcJmmDoiQHk6pxS90DbC5pE0mrA4cCV/d1fs3MrHU0VOvZ7I+ShgOLgRMj4gVJ/ydpO1J161PA8QCSRgHnR8QBEbFE0meBG4B+wIUR8VB9foJV4sc6mVkzU8QKl/1axvjx46O9vb2my2zloFH6WCdIXaCdccjWLbMOGkErb4PWOyTdGxHj652Pemi46tlVWSFozFywiODNZwFOmjyz3lmrCT/Wqf5afRs06ykHzRpq9aDhxzrVX6tvg2Y95aBZQ60eNPxYp/pr9W3QrKccNGuo1YOGH+tUf62+DZr1lINmDbV60Jiw/WjOOGRrRg9pQ8DoIW1uBFRjrb4NmvVUI95yssryswD9WKd68zZo1jO+5aTGt5w0O9+uYGatfMuJS5pWtdL7LAu3KwAOnGbWEnxN06rm2xXMrNU5aFrVfLuCmbU6V89a1UYNaWNmmQDp2xW6x9eFzZqXS5pWNd+u0HPuxs6suTloWtV8n2XP+bqwWXNz9ax1i++z7BlfFzZrbi5pmtWQu7Eza24Omt10z9W/4rnTNmPZNwfz3Gmbcc/Vv6p3lrpnyhXww63gtCHpfcoV9c5RS/F1YbPm5urZbrjn6l+x1b3/TZteB8H6dDD43v/mHmDHDx5f7+x1bcoV8OfPw+JcFfjis+k7wDYfrV++Woi7sTNrbg3VjZ6kLwCfAgScFxE/knQ2cCDwOvA4cHRELCgz7VPAy8BSYEk1XTx1txu9507bjPXpWHE4I1n/tMeqnk/d/HCrFChLDd4QvjS19vkxs6bUyt3oNUz1rKStSAFzJ2Bb4AOSNgduAraKiG2AfwOndDKb90XEdn31Z64bKwbMNHxeXyyu9704o3vDzcxsOQ0TNIEtgbsiYmFELAFuBw6OiBvzd4C7gDH1yuBcjawwfESNc7KSBq1XYUTApR+FedNrmh0zs2bTSEFzKrCbpOGSBgIHABuWpPkkcF2F6QO4UdK9ko6rtBBJx0lql9Te0VG+5FjJsztMZFGsvtyw16M/z+4wsVvzqYuXn4Mlr684fEAbbPVheOZO+PnOcN3JsHB+7fNnZtYEGiZoRsQjwFmk6tjrgQeAQgkTSV/P3y+tMItdI2IHYH/gREm7VVjOuRExPiLGjxxZvuRYyY4fPJ6p7/guz6lOLY0AABUmSURBVDGSZQFLWI3XB67PjgdWjNGN4bUX4ZIPwZL/wPu+nq5hovR+4E/gwxfA5+6D7T8O//oV/HQHuPtcWLq43jk3M2soDdUQqJik04EZEfFzSUcCJwB7RsTCKqY9DXglIn7QWboeP0/znvPhmq/AUdfA2Hev/Hz60uLXUsB89m44/ArYdI/O0z83FW44BZ68A0aMg31Ph833qk1eq+S+W83qyw2BGoSkdfP7RsAhwGWS9gNOAj5YKWBKWkvSoMJnYB9SdW/f2u5wGDgc/vnTPl/USlm2FK48Fp7+Oxz8y64DJsD6W8EnroZDL4Nli+HSD8ElH4aOf/d9fqvgvlvNrJ4aKmgCf5T0MPBn4MSIeAH4GTAIuEnS/ZJ+CSBplKRr83TrAX+X9ADwL+CaiLi+z3M7oA12Og7+fT3MfbTPF9ctEakU/MifYb8zYesPVz+tBFscAJ+5G/b5biql/nxnuO4kaL+wrp0juO9WM6unhurcICLeU2bYZhXSziI1FiIiniDdplJ7Ox4Lf/8h3PkzOOhndclCWbefBff+Gt79Jdj50ys3j/6rw7s+B9seBrd+D+7+5fLj69A5gvtuNbN6arSSZvNZa0Sqpp1yeWqh2gjuuQBuOwO2OwL2/GbP57fWCPjAD2HtMresLF4EN/53+Za5fcB9t5pZPTlo9oZdTkwtTe9ugH5oH74qVcu+dT848MepqrW3vDK3wvA5cPoo+OV74KoT4V/nwTN3w+uvrpi2h33fuu9WM6unhqqebVrDN4UtPwDtF8B7vgJrrF2ffDz5N/jjsTBmR/jwr6FfL/+9g8eU74Zv4PB0u8rsB2DadTD5kjxCMGJz2GBbWH8bWPQC3PULWLLyfd+671Yzq6eGveWkFnp8y0mxZ++BC/ZKjW5W9hpiT8yeAhe9HwZtAJ+8HgYO6/1llHb4Dqkx1IE/eTPoRcBLM1N+Zj8Az+X3lzpp3eq+b82aSivfcuKSZm/ZcEfYcGe48+ew46d6v5TXmflPwqUfhjUGwcev7JuACW8Gxpu/nfqrHTwG9jx1+VKilIYPHpNa4Ba8Og/O3ozUcVMJ931rZk3C1zR7066fhxefgYcn1W6Zr3TAJYek3n6OuDIFq760zUdTqfC0Bem92laza43oJG+RroU6eJpZg3PQ7E1v3R+Gbwb//Emqpuxr/3k5lTBfmg2H/x7W3aLvl9kTe56aqnOL9W+DTfdOVb8/2QFu/Ib7vjWzhuXq2d602mqwy2fhL1+Ep/4Gm5Tt/rZnplzxZvVo/9XTrR6H/Q423Kn3l9XbOqveXfAM3Hp66l3pvovT/aXvPGHFIGtmVkduCNRbDYEKFr8GP9oKNtgOjvhD7867XEOcfqvDQf9bs84F+tych+Cv34LpN8CgUbD7yek+2FpeI+6C+761VtfKDYFcPdvbBqyZutZ77CaY83Dvzvvmby8fMAGWvp6GryrWe3vqWP6oa2Hw6HSS8ItdUneADXCC575vzVqbg2Zf2PFYGDAwda3Xmyo1lFkVG9CM3RWOuQn+X34S3OVHwAV7w1N/73EHCT3hvm/NWpuDZl8YOAy2PyIdzF+a3XvzbRtSfnhft5itFyl1GvHpO+GDP4UXZ6Z7Uf90fO5kId7sIKFGgdN935q1NgfNvrLzZyCWrtjJ+cqadj0sWgAq+csGtKXGNKuyfv1hh0/A5++DNQdDLFt+/OJFNauidt+3PTdp8kx2PfMWNjn5GnY98xZXbVtTcdDsK8M2gS0/CO2/TreG9MQzd8Pvj0rd0R3449SDDkrvxb3xrOoGtMFrL5UfV6Mqavd92zO+JmzNrnGaJK6K3vX51NHBfb9JnbqvjLmPwm8/CutsAIf/AdYemUpdrapS/7flnsDSB9z3bc90dk3Y69CagYNmXxrzDth419S13k7HQb8B3Zv+xRmpt5/+a8DH/5QCZqvb89QVb7uB1Bn8o9fAFu/vchY9vWVkwvajfYBfSb4mbM3O1bN97V2fh5dmwEPd7Fpv4Xy45EOpOvLwP8DQsX2SvaazzUdTlXRxFfV+Z8F6b4PffQxu+R4sW1ZxclcP1pevCVuzc9Dsa5vvAyPeCv/8cfX3Gb6+EC47FOY/AYf9FjbYpm/z2GxK+7/d+QQ4+vr00O07vp/W3aIFZSf1LSP15WvC1uwaLmhK+oKkqZIekvTFPGyYpJskTc/vQytMe2ROM13SkbXNeQWrrQbv+hw89yA8eXvX6ZcugT8cDc/+Cw45r2+64lsVDVgTDvoZvP8cePxmOO99MPeRFZK5erC+Jmw/mjMO2ZrRQ9oQMHpIG2ccsnVzVXfX8T5hq7+G6kZP0lbA74CdgNeB64FPA58C5kfEmZJOBoZGxEkl0w4D2oHxpOdP3Qu8IyJeqLS8PulGr5wl/0k71/pbp0d3VRIBV382PcT5/eekThKs+56+E674BLz+Kkz4Obx9whujdj3zFmaWCZCjh7Txj5P3qGUum1dx/8flHg+3Ki+/mmfKtgB3o9c4tgTuioiFEbEEuB04GDgIuDinuRiYUGbafYGbImJ+DpQ3AfvVIM9d678GvPP4VAJ6rpOHLd/ynRQwd/svB8ye2HgXOP72dJ3z90fCTd+EZalK1tWDPVQIGnXqXKJXlr+yJcVly9K2VNoIrYb3CVv9NVrr2anA9yQNBxYBB5BKj+tFxGyAiJgtad0y044Giu9FmJGHLUfSccBxABtttFHv5r4z4z8Jf/uf1LXewWU6PLj7V/C3c2CHI+F9X6tdvlZV64yCo66B606Cf/wInpsCH7rAt4z0xLJlcNOp5YPGX0+rvqTVnZLismXw8ix4/nF4/rHKQesvX4aXZsLA4TBwRHp+68Dh6X2NdVLvUoVlF5cUC0E3At7y3jSPF2fCS7NSA76XZr35/eVZsGxJ+Xy+OCPNo7AcW2U1VPUsgKRjgBOBV4CHScHz6IgYUpTmhYgYWjLdRGCNiPhu/v4NYGFEnFNpWTWrni247iS453z4wpTUGXnB1D/CH45Jt0t85OKGeqLHKuHei+Har8KgDeDQS9O1znpWL/ZUT6snq5n+9VfTAweemwJzpqYakjkPweJXK8+3bRgM2ajya41Blas39/p2qhl4/nGY/3h+fyK9lrzWvfVTarUBbwbQedNh6X+qm67fGunka/CY9L7OaGi/EF4r38iMQaNg3P6wxQEwdrf06L5VVCtXzzZc0Cwm6XRSifELwO65lLkBcFtEjCtJe1hOc3z+/quc7rJK86950HzhafjJ9qmjg32+k4Y9fitc+hEYs2O63unnR/aNGe1w+cfhlQ5YTenpMAXNdE2qp9fUyk3ff810OaBtSGqw9tzUFKzIx4Y1BsP6+Zr8lMvTPbGl1hwMW30oPRe18CoNdm1D4T+vwLLFnedxtQGpR61hm8Lw/Cp8vnDf8r0/Dd4QTrwbXp0HC59Pr1fnwcJ5yw+bdm3l5R7wgxQYB49O7wOHr1hyLLv+2mDbQ+HVDnj8Fli8EFYfBJvvBePeD5vvvXy/0bU46enL6XHQrHceliNp3YiYK2kj4EZgF+BrwPNFDYGGRcR/lUw3jNT4Z4c86D5SQ6D5lZZV86AJcP5e6QAOsPa66daI4ZvB0ddW7pDdescrc9M1rHIljcEbpttXGlkE/OCt8OrcFcdptfQb+q+RSkj9Vy//Pu3adFCvZOgmKUCut/WbgXLwhpWrN6F80I5IQWTBM7DgaVjwbPrcfkHlZR9xZQqMgzeE1fqVT9PTk4YfblW+R6nu/P+dBZ3Fi+CJ22HaNTDturQOVuufOjnZ4v2p3+Sbv9W7Jz21nD5r5aDZiPWAf8zXNBcDJ0bEC5LOBK7IVbfPAB8BkDQeOCEijo2I+ZK+A9yT5/PtzgJmXUy5IlV3Fc7gX5kDCN5xlANmLay97vIlzGIvzkitnPuvUds8VeOVufDAZak7xnIBE9LBeKOd029Y+vqb768vhKUvwJLX08lCxYApOPkZWHOdzvNSOLB2VVKR0vpee10YU3RsnX5j5aC12Z6dL7s7y6+kXI9S3X3owTYfrby8AW0wbr/0WrYMZrannqqmXQvX/Vf5aRYvgmsnphJ8RH4gQX4v/f73H1e4pvuldDK+bHG67rp0SXpftiQPW5o+P3l72jZKp7/5281R09IAGq6kWUs1L2n2xlmu9Uyl/wDSM1DHvhs23RM23QNGbF6/hh3LlqaqvvsuTiWWZUtgo12gYxosKnMuWO02VO9tsBFu2ajXLSvzHoOfvaPv5r/mkFSq7Tcgva/WL1V1r9Y/D+8PsyZXmFips5AquaRptdFKD5FuVOVKGv3bYMdj0hn447ek0hCkQLLpHun1lvema3LQtwfdBc+k244mX/Jma9B3npA66R85rnLQqbak1BslrZ7oaUmxt/JQj1LViM3SNlXupGWdUXD839NJmpQfAZjfi7//7B2Vr+n26KRpFX0mbx9w0KylSk/o8AZbO9UctF94KgXPx26Gh/6USntaDUa/A9ZePwXVwnXRwi0LxfPuSmnQfd/XUin3vt+k5UIK1PueDuMOWL4VZk+DTisHrUZQ6aRlr2/BWsOrmP6bzX3StApw9Wwtq2cboWrKumfpknRd6rGbU0CbWWF7GTAw3YvbNhQGDku3XxTeC8MGtJXfBgrWGQPbHwHbH55u0bBVU71bv7r1bI84aNa69Wy9uyCznjltCG805Co1YGDnLVP7t6XGObF0xXEDR8BX/1251ahZA2nloOnq2Vpr5aqpVUHFKvZ8TWnxotQKcuH81GBn4fz0vfD5zp+Vn+/C5x0wzZqAg6ZZd3R1TWhAW3qtM6r89A9f5evaZk2s0TpsN2ts5R6C3Z1r0nueumKvT26IYdY0XNI0666eVLE3QutVM1tpDppmtebr2mZNy0HTzKyGJk2e6UfTNTEHTTOzGpk0eSanXPkgixan245mLljEKVc+CODA2STcEMjMrEbOvmHaGwGzYNHipZx9w7Q65ci6y0HTzKxGZi0o0xNUJ8Ot8ThompnVyKgh5R8yX2m4NR4HTTOzGpm47zjaBizf81PbgH5M3HdcnXJk3eWGQGZmNVJo7OPWs83LQdPMrIYmbD/aQbKJNVzQlPQl4FjSoyQeBI4GbgIG5STrAv+KiAllpl2apwF4JiI+2Pc5NjOzVtFQQVPSaODzwNsiYpGkK4BDI+I9RWn+CFxVYRaLImK7GmTVzMxaUCM2BOoPtEnqDwwEZhVGSBoE7AFMqlPezMyshTVU0IyImcAPgGeA2cCLEXFjUZKDgZsj4qUKs1hTUrukuyStUH0LIOm4nKa9o6OjV/NvZmartoYKmpKGAgcBmwCjgLUkHVGU5DDgsk5msVF+mvjHgB9J2rQ0QUScGxHjI2L8yJEjezH3Zma2qmuoa5rAXsCTEdEBIOlK4F3AJZKGAzuRSptlRcSs/P6EpNuA7YHHK6W/995750l6uvey36tGAPPqnYlONHr+oPHz6Pz1jPPXMz3J38a9mZFm0mhB8xlgZ0kDgUXAnkB7HvcR4C8R8Vq5CXMpdWFE/EfSCGBX4PudLSwiGraoKak9l5obUqPnDxo/j85fzzh/PdPo+WtUDVU9GxF3A38A7iPdOrIacG4efSglVbOSxks6P3/dEmiX9ABwK3BmRDxck4ybmVlLaLSSJhHxTeCbZYbvXmZYO+meTiLin8DWfZ0/MzNrXQ1V0rTlnNt1krpq9PxB4+fR+esZ569nGj1/DUkRUe88mJmZNQWXNM3MzKrkoGlmZlYlB806krShpFslPSLpIUlfKJNmd0kvSro/v06tcR6fkvRgXnZ7mfGS9BNJj0maImmHGuZtXNF6uV/SS5K+WJKm5utP0oWS5kqaWjRsmKSbJE3P70MrTHtkTjNd0pE1zN/Zkh7N/+GfJA2pMG2n20Mf5u80STOL/scDKky7n6RpeXs8uYb5u7wob09Jur/CtH26/iodUxpp+2t6EeFXnV7ABsAO+fMg4N+kzuqL0+xOuj+1Xnl8ChjRyfgDgOsAATsDd9cpn/2A54CN673+gN2AHYCpRcO+D5ycP58MnFVmumHAE/l9aP48tEb52wfonz+fVS5/1WwPfZi/04CvVrENPA68BVgdeKB0f+qr/JWMPwc4tR7rr9IxpZG2v2Z/uaRZRxExOyLuy59fBh4Bmu1BewcBv4nkLmCIpA3qkI89gccjou49PEXEHcD8ksEHARfnzxcD5fpG3he4KSLmR8QLpEfi7VeL/EXEjRGxJH+9CxjT28utVoX1V42dgMci4omIeB34HWm996rO8idJwEfpvLvPPtPJMaVhtr9m56DZICSNJXX7d3eZ0btIekDSdZLeXtOMpeea3ijpXknHlRk/Gni26PsM6hP4V+j8okg911/BehExG9KBjfRc2FKNsi4/Sao9KKer7aEvfTZXH19YoXqxEdbfe4A5ETG9wviarb+SY0ozbX8NzUGzAUhaG/gj8MVY8Qku95GqHLcFfkrtH4u2a0TsAOwPnChpt5LxKjNNTe9jkrQ68EHg92VG13v9dUcjrMuvA0uASysk6Wp76Cu/ADYFtiM9AemcMmnqvv7o+qESNVl/XRxTKk5WZpjvSSzhoFlnkgaQNu5LI+LK0vER8VJEvJI/XwsMUOpbtybizU7w5wJ/IlWBFZsBbFj0fQxFz0Ctkf2B+yJiTumIeq+/InMK1db5fW6ZNHVdl7nhxweAwyNf5CpVxfbQJyJiTkQsjYhlwHkVllvv9dcfOAS4vFKaWqy/CseUht/+moWDZh3l6x8XAI9ExP9USLN+ToeknUj/2fM1yt9aSg/+RtJapMYiU0uSXQ18Irei3Zn0DNTZtchfkYpn9/VcfyWuBgqtEY8EriqT5gZgH0lDc/XjPnlYn5O0H3AS8MGIWFghTTXbQ1/lr/g6+cEVlnsPsLmkTXLtw6Gk9V4rewGPRsSMciNrsf46OaY09PbXVOrdEqmVX8C7SdUfU4D78+sA4ATghJzms8BDpJaAdwHvqmH+3pKX+0DOw9fz8OL8CfhfUqvFB4HxNV6HA0lBcHDRsLquP1IAnw0sJp29HwMMB24Gpuf3YTnteOD8omk/CTyWX0fXMH+Pka5nFbbDX+a0o4BrO9seapS//8vb1xRSANigNH/5+wGkFqOP1zJ/efhFhe2uKG1N118nx5SG2f6a/eVu9MzMzKrk6lkzM7MqOWiamZlVyUHTzMysSg6aZmZmVXLQNDMzq5KDppmZWZUcNM3MzKr0/wH+N5fpt+rF5gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot score vs number of ensemble members\n",
    "print('RMSE Single Learners %.3f (%.3f)' % (mean(single_mse), std(single_mse)))\n",
    "print('RMSE Ensemble Learners %.3f (%.3f)' % (mean(ensemble_mse), std(ensemble_mse)))\n",
    "x_axis = [i for i in range(1, n_splits+1)]\n",
    "pyplot.plot(x_axis, single_mse, marker='o', linestyle='None')\n",
    "pyplot.plot(x_axis, ensemble_mse, marker='o')\n",
    "plt.title(\"Root Mean Squared Error: Single Learner vs Ensemble Learners (Bagging)\")\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> 1: single=48.784, ensemble=48.784\n",
      "> 2: single=61.850, ensemble=50.509\n",
      "> 3: single=64.474, ensemble=53.655\n",
      "> 4: single=72.977, ensemble=57.670\n",
      "> 5: single=53.740, ensemble=54.455\n",
      "> 6: single=52.101, ensemble=53.326\n",
      "> 7: single=50.242, ensemble=51.322\n",
      "> 8: single=53.941, ensemble=51.090\n",
      "> 9: single=49.958, ensemble=49.955\n",
      "> 10: single=48.491, ensemble=49.166\n",
      "> 11: single=55.887, ensemble=49.453\n",
      "> 12: single=52.083, ensemble=48.977\n",
      "> 13: single=69.410, ensemble=49.857\n",
      "> 14: single=52.819, ensemble=49.827\n",
      "> 15: single=61.754, ensemble=50.255\n",
      "> 16: single=48.724, ensemble=49.553\n",
      "> 17: single=53.619, ensemble=49.122\n",
      "> 18: single=48.566, ensemble=48.444\n",
      "> 19: single=52.491, ensemble=48.013\n",
      "> 20: single=54.934, ensemble=47.951\n"
     ]
    }
   ],
   "source": [
    "single_mae, ensemble_mae = list(), list()\n",
    "for i in range(1, n_splits+1):\n",
    "    subset = members[:i]\n",
    "    yhats = [model.predict(X_test) for model in subset]\n",
    "    yhats = array(yhats)\n",
    "    averaged = np.average(yhats, axis=0)\n",
    "    ensemble = mean_absolute_error(test_target,averaged)\n",
    "    \n",
    "    *_, y1 = [members[i-1].predict(X_test)]\n",
    "    single =  mean_absolute_error(test_target,y1)\n",
    "    print('> %d: single=%.3f, ensemble=%.3f' % (i, single, ensemble))\n",
    "    single_mae.append(single)\n",
    "    ensemble_mae.append(ensemble)\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE Single Learners 55.342 (6.907)\n",
      "MAE Ensemble Learners 50.569 (2.398)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot score vs number of ensemble members\n",
    "print('MAE Single Learners %.3f (%.3f)' % (mean(single_mae), std(single_mae)))\n",
    "print('MAE Ensemble Learners %.3f (%.3f)' % (mean(ensemble_mae), std(ensemble_mae)))\n",
    "x_axis = [i for i in range(1, n_splits+1)]\n",
    "pyplot.plot(x_axis, single_mae, marker='o', linestyle='None')\n",
    "pyplot.plot(x_axis, ensemble_mae, marker='o')\n",
    "plt.title(\"Mean Absolute Error: Single Learner vs Ensemble Learners (Bagging)\")\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> 1: single=729.729, ensemble=729.729\n",
      "> 2: single=705.886, ensemble=717.808\n",
      "> 3: single=701.848, ensemble=712.488\n",
      "> 4: single=670.617, ensemble=702.020\n",
      "> 5: single=745.110, ensemble=710.638\n",
      "> 6: single=740.329, ensemble=715.586\n",
      "> 7: single=726.111, ensemble=717.090\n",
      "> 8: single=702.343, ensemble=715.247\n",
      "> 9: single=773.521, ensemble=721.722\n",
      "> 10: single=758.062, ensemble=725.356\n",
      "> 11: single=738.630, ensemble=726.562\n",
      "> 12: single=763.960, ensemble=729.679\n",
      "> 13: single=673.506, ensemble=725.358\n",
      "> 14: single=742.615, ensemble=726.590\n",
      "> 15: single=707.894, ensemble=725.344\n",
      "> 16: single=817.977, ensemble=731.134\n",
      "> 17: single=769.569, ensemble=733.394\n",
      "> 18: single=768.523, ensemble=735.346\n",
      "> 19: single=796.741, ensemble=738.577\n",
      "> 20: single=771.720, ensemble=740.234\n"
     ]
    }
   ],
   "source": [
    "single_maxe, ensemble_maxe = list(), list()\n",
    "for i in range(1, n_splits+1):\n",
    "    subset = members[:i]\n",
    "    yhats = [model.predict(X_test) for model in subset]\n",
    "    yhats = array(yhats)\n",
    "    averaged = np.average(yhats, axis=0)\n",
    "    ensemble = max_error(test_target,averaged)\n",
    "    \n",
    "    *_, y1 = [members[i-1].predict(X_test)]\n",
    "    single =  max_error(test_target,y1)\n",
    "    print('> %d: single=%.3f, ensemble=%.3f' % (i, single, ensemble))\n",
    "    single_maxe.append(single)\n",
    "    ensemble_maxe.append(ensemble)\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max Error Single Learners 740.235 (37.851)\n",
      "Max Error Ensemble Learners 723.995 (9.645)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot score vs number of ensemble members\n",
    "print('Max Error Single Learners %.3f (%.3f)' % (mean(single_maxe), std(single_maxe)))\n",
    "print('Max Error Ensemble Learners %.3f (%.3f)' % (mean(ensemble_maxe), std(ensemble_maxe)))\n",
    "x_axis = [i for i in range(1, n_splits+1)]\n",
    "pyplot.plot(x_axis, single_maxe, marker='o', linestyle='None')\n",
    "pyplot.plot(x_axis, ensemble_maxe, marker='o')\n",
    "plt.title(\"Max Error: Single Learner vs Ensemble Learners (Bagging)\")\n",
    "pyplot.show()    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> 1: single=0.926, ensemble=0.926\n",
      "> 2: single=0.929, ensemble=0.930\n",
      "> 3: single=0.920, ensemble=0.928\n",
      "> 4: single=0.921, ensemble=0.927\n",
      "> 5: single=0.923, ensemble=0.927\n",
      "> 6: single=0.921, ensemble=0.927\n",
      "> 7: single=0.927, ensemble=0.928\n",
      "> 8: single=0.933, ensemble=0.930\n",
      "> 9: single=0.928, ensemble=0.930\n",
      "> 10: single=0.925, ensemble=0.930\n",
      "> 11: single=0.911, ensemble=0.929\n",
      "> 12: single=0.924, ensemble=0.929\n",
      "> 13: single=0.913, ensemble=0.929\n",
      "> 14: single=0.928, ensemble=0.929\n",
      "> 15: single=0.922, ensemble=0.929\n",
      "> 16: single=0.923, ensemble=0.929\n",
      "> 17: single=0.914, ensemble=0.929\n",
      "> 18: single=0.933, ensemble=0.929\n",
      "> 19: single=0.922, ensemble=0.929\n",
      "> 20: single=0.918, ensemble=0.929\n"
     ]
    }
   ],
   "source": [
    "single_evs, ensemble_evs = list(), list()\n",
    "for i in range(1, n_splits+1):\n",
    "    subset = members[:i]\n",
    "    yhats = [model.predict(X_test) for model in subset]\n",
    "    yhats = array(yhats)\n",
    "    averaged = np.average(yhats, axis=0)\n",
    "    ensemble = explained_variance_score(test_target,averaged)\n",
    "    \n",
    "    *_, y1 = [members[i-1].predict(X_test)]\n",
    "    single =  explained_variance_score(test_target,y1)\n",
    "    print('> %d: single=%.3f, ensemble=%.3f' % (i, single, ensemble))\n",
    "    single_evs.append(single)\n",
    "    ensemble_evs.append(ensemble)\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Explained Variance Score Single Learners 0.923 (0.006)\n",
      "Explained Variance Score Ensemble Learners 0.929 (0.001)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot score vs number of ensemble members\n",
    "print('Explained Variance Score Single Learners %.3f (%.3f)' % (mean(single_evs), std(single_evs)))\n",
    "print('Explained Variance Score Ensemble Learners %.3f (%.3f)' % (mean(ensemble_evs), std(ensemble_evs)))\n",
    "x_axis = [i for i in range(1, n_splits+1)]\n",
    "pyplot.plot(x_axis, single_evs, marker='o', linestyle='None')\n",
    "pyplot.plot(x_axis, ensemble_evs, marker='o')\n",
    "plt.title(\"Explained Variance Score Single Learners: Single Learner vs Ensemble Learners (Bagging)\")\n",
    "pyplot.show()    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Conclusion**\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; Ensemble learning of PCA pre-processed stacked long-short term memory was used to predict solar irradiance using weather parameters as its features. Stacked long-short term memory was used as base learners since it is suitable for inputs such as time series sequence. It also overcome problems such as vanishing gradient problem. Based from the plots that shows the performance comparison of base learners and ensemble learners, it shows that at some iteration, ensemble learners get stable and consistently improves the prediction in terms of metrics used such as mean absolute error, root mean squared error, maximum residual error, explained variance score, and regression score function. </p>\n",
    "\n",
    "<p style='text-align: justify;'>&emsp; It is recommended to use other ensemble learning methods such as stacking or parallel grid search algorithms for prediction problems. Also, it is recommended to add concentration of dust particles as features to consider possible power attenuation. </p>"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}