ml-finance-python

python scripts for finance machine learning

git clone https://9o.is/git/ml-finance-python.git

03_multivariate_timeseries.ipynb

(173828B)


      1 {
      2  "cells": [
      3   {
      4    "cell_type": "markdown",
      5    "metadata": {},
      6    "source": [
      7     "# Multivariate Time Series Regression"
      8    ]
      9   },
     10   {
     11    "cell_type": "markdown",
     12    "metadata": {},
     13    "source": [
     14     "So far, we have limited our modeling efforts to single time series. RNNs are naturally well suited to multivariate time series and represent a non-linear alternative to the Vector Autoregressive (VAR) models we covered in Chapter 8, Time Series Models. "
     15    ]
     16   },
     17   {
     18    "cell_type": "markdown",
     19    "metadata": {},
     20    "source": [
     21     "## Run inside docker container for GPU acceleration"
     22    ]
     23   },
     24   {
     25    "cell_type": "markdown",
     26    "metadata": {},
     27    "source": [
     28     "See [tensorflow guide](https://www.tensorflow.org/install/docker) and more detailed [instructions](https://blog.sicara.com/tensorflow-gpu-opencv-jupyter-docker-10705b6cd1d)"
     29    ]
     30   },
     31   {
     32    "cell_type": "markdown",
     33    "metadata": {},
     34    "source": [
     35     "`docker run -it -p 8889:8888 -v /path/to/machine-learning-for-trading/18_recurrent_neural_nets:/rnn --name tensorflow tensorflow/tensorflow:latest-gpu-py3 bash`"
     36    ]
     37   },
     38   {
     39    "cell_type": "markdown",
     40    "metadata": {},
     41    "source": [
     42     "Inside docker container: \n",
     43     "`jupyter notebook --ip 0.0.0.0 --no-browser --allow-root`"
     44    ]
     45   },
     46   {
     47    "cell_type": "markdown",
     48    "metadata": {},
     49    "source": [
     50     "## Imports & Settings"
     51    ]
     52   },
     53   {
     54    "cell_type": "code",
     55    "execution_count": 59,
     56    "metadata": {},
     57    "outputs": [],
     58    "source": [
     59     "%matplotlib inline\n",
     60     "import numpy as np\n",
     61     "import pandas as pd\n",
     62     "import matplotlib.pyplot as plt\n",
     63     "import seaborn as sns\n",
     64     "import pandas_datareader.data as web\n",
     65     "from datetime import datetime, date\n",
     66     "from sklearn.metrics import mean_squared_error, mean_absolute_error\n",
     67     "from sklearn.preprocessing import minmax_scale\n",
     68     "from keras.callbacks import ModelCheckpoint, EarlyStopping\n",
     69     "from keras.models import Sequential, Model\n",
     70     "from keras.layers import Dense, LSTM, Input, concatenate, Embedding, Reshape\n",
     71     "import keras\n",
     72     "import keras.backend as K\n",
     73     "import tensorflow as tf"
     74    ]
     75   },
     76   {
     77    "cell_type": "code",
     78    "execution_count": 2,
     79    "metadata": {},
     80    "outputs": [],
     81    "source": [
     82     "sns.set_style('whitegrid')\n",
     83     "np.random.seed(42)\n",
     84     "K.clear_session()"
     85    ]
     86   },
     87   {
     88    "cell_type": "markdown",
     89    "metadata": {},
     90    "source": [
     91     "## Load Data"
     92    ]
     93   },
     94   {
     95    "cell_type": "markdown",
     96    "metadata": {},
     97    "source": [
     98     "For comparison, we illustrate the application of RNNs to modeling and forecasting several time series using the same dataset we used for the VAR example, monthly data on consumer sentiment, and industrial production from the Federal Reserve's FRED service in Chapter 8, Time Series Models:"
     99    ]
    100   },
    101   {
    102    "cell_type": "code",
    103    "execution_count": 3,
    104    "metadata": {},
    105    "outputs": [
    106     {
    107      "name": "stdout",
    108      "output_type": "stream",
    109      "text": [
    110       "<class 'pandas.core.frame.DataFrame'>\n",
    111       "DatetimeIndex: 456 entries, 1980-01-01 to 2017-12-01\n",
    112       "Data columns (total 2 columns):\n",
    113       "sentiment    456 non-null float64\n",
    114       "ip           456 non-null float64\n",
    115       "dtypes: float64(2)\n",
    116       "memory usage: 10.7 KB\n"
    117      ]
    118     }
    119    ],
    120    "source": [
    121     "df = web.DataReader(['UMCSENT', 'IPGMFN'], 'fred', '1980', '2017-12').dropna()\n",
    122     "df.columns = ['sentiment', 'ip']\n",
    123     "df.info()"
    124    ]
    125   },
    126   {
    127    "cell_type": "code",
    128    "execution_count": 4,
    129    "metadata": {},
    130    "outputs": [
    131     {
    132      "data": {
    133       "text/html": [
    134        "<div>\n",
    135        "<style scoped>\n",
    136        "    .dataframe tbody tr th:only-of-type {\n",
    137        "        vertical-align: middle;\n",
    138        "    }\n",
    139        "\n",
    140        "    .dataframe tbody tr th {\n",
    141        "        vertical-align: top;\n",
    142        "    }\n",
    143        "\n",
    144        "    .dataframe thead th {\n",
    145        "        text-align: right;\n",
    146        "    }\n",
    147        "</style>\n",
    148        "<table border=\"1\" class=\"dataframe\">\n",
    149        "  <thead>\n",
    150        "    <tr style=\"text-align: right;\">\n",
    151        "      <th></th>\n",
    152        "      <th>sentiment</th>\n",
    153        "      <th>ip</th>\n",
    154        "    </tr>\n",
    155        "    <tr>\n",
    156        "      <th>DATE</th>\n",
    157        "      <th></th>\n",
    158        "      <th></th>\n",
    159        "    </tr>\n",
    160        "  </thead>\n",
    161        "  <tbody>\n",
    162        "    <tr>\n",
    163        "      <th>1980-01-01</th>\n",
    164        "      <td>67.0</td>\n",
    165        "      <td>46.8853</td>\n",
    166        "    </tr>\n",
    167        "    <tr>\n",
    168        "      <th>1980-02-01</th>\n",
    169        "      <td>66.9</td>\n",
    170        "      <td>47.9806</td>\n",
    171        "    </tr>\n",
    172        "    <tr>\n",
    173        "      <th>1980-03-01</th>\n",
    174        "      <td>56.5</td>\n",
    175        "      <td>48.4758</td>\n",
    176        "    </tr>\n",
    177        "    <tr>\n",
    178        "      <th>1980-04-01</th>\n",
    179        "      <td>52.7</td>\n",
    180        "      <td>47.0631</td>\n",
    181        "    </tr>\n",
    182        "    <tr>\n",
    183        "      <th>1980-05-01</th>\n",
    184        "      <td>51.7</td>\n",
    185        "      <td>45.6939</td>\n",
    186        "    </tr>\n",
    187        "  </tbody>\n",
    188        "</table>\n",
    189        "</div>"
    190       ],
    191       "text/plain": [
    192        "            sentiment       ip\n",
    193        "DATE                          \n",
    194        "1980-01-01       67.0  46.8853\n",
    195        "1980-02-01       66.9  47.9806\n",
    196        "1980-03-01       56.5  48.4758\n",
    197        "1980-04-01       52.7  47.0631\n",
    198        "1980-05-01       51.7  45.6939"
    199       ]
    200      },
    201      "execution_count": 4,
    202      "metadata": {},
    203      "output_type": "execute_result"
    204     }
    205    ],
    206    "source": [
    207     "df['1980':].head()"
    208    ]
    209   },
    210   {
    211    "cell_type": "code",
    212    "execution_count": 5,
    213    "metadata": {},
    214    "outputs": [
    215     {
    216      "data": {
    217       "image/png": "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\n",
    218       "text/plain": [
    219        "<Figure size 432x288 with 1 Axes>"
    220       ]
    221      },
    222      "metadata": {},
    223      "output_type": "display_data"
    224     }
    225    ],
    226    "source": [
    227     "df.plot();"
    228    ]
    229   },
    230   {
    231    "cell_type": "markdown",
    232    "metadata": {},
    233    "source": [
    234     "## Prepare Data"
    235    ]
    236   },
    237   {
    238    "cell_type": "markdown",
    239    "metadata": {},
    240    "source": [
    241     "### Scaling"
    242    ]
    243   },
    244   {
    245    "cell_type": "markdown",
    246    "metadata": {},
    247    "source": [
    248     "First we scale the data to the [0,1] interval:"
    249    ]
    250   },
    251   {
    252    "cell_type": "code",
    253    "execution_count": 6,
    254    "metadata": {},
    255    "outputs": [],
    256    "source": [
    257     "df_scaled = df.apply(minmax_scale)"
    258    ]
    259   },
    260   {
    261    "cell_type": "markdown",
    262    "metadata": {},
    263    "source": [
    264     "### Stationarity"
    265    ]
    266   },
    267   {
    268    "cell_type": "markdown",
    269    "metadata": {},
    270    "source": [
    271     "We apply the same transformation—annual difference for both series, prior log-transform for industrial production—to achieve stationarity that we used in Chapter 8 on Time Series Models:"
    272    ]
    273   },
    274   {
    275    "cell_type": "code",
    276    "execution_count": 35,
    277    "metadata": {},
    278    "outputs": [],
    279    "source": [
    280     "df_transformed = pd.DataFrame({'ip': np.log(df.ip).diff(12),\n",
    281     "                              'sentiment': df.sentiment.diff(12)}).dropna()"
    282    ]
    283   },
    284   {
    285    "cell_type": "markdown",
    286    "metadata": {},
    287    "source": [
    288     "We can reshape directly to get non-overlapping series, i.e., one sample for each year (works only if the number of samples is divisible by window size):"
    289    ]
    290   },
    291   {
    292    "cell_type": "code",
    293    "execution_count": 7,
    294    "metadata": {},
    295    "outputs": [
    296     {
    297      "data": {
    298       "text/plain": [
    299        "(38, 12, 2)"
    300       ]
    301      },
    302      "execution_count": 7,
    303      "metadata": {},
    304      "output_type": "execute_result"
    305     }
    306    ],
    307    "source": [
    308     "df_scaled.values.reshape(-1, 12, 2).shape"
    309    ]
    310   },
    311   {
    312    "cell_type": "markdown",
    313    "metadata": {},
    314    "source": [
    315     "### Reshape data into RNN format"
    316    ]
    317   },
    318   {
    319    "cell_type": "markdown",
    320    "metadata": {},
    321    "source": [
    322     "The create_multivariate_rnn_data function transforms a dataset of several time series into the shape required by the Keras RNN layers, namely `n_samples` x `window_size` x `n_series`, as follows:"
    323    ]
    324   },
    325   {
    326    "cell_type": "code",
    327    "execution_count": 9,
    328    "metadata": {},
    329    "outputs": [],
    330    "source": [
    331     "def create_multivariate_rnn_data(data, window_size):\n",
    332     "    y = data[window_size:]\n",
    333     "    n = data.shape[0]\n",
    334     "    X = np.stack([data[i: j] for i, j in enumerate(range(window_size, n))], axis=0)\n",
    335     "    return X, y"
    336    ]
    337   },
    338   {
    339    "cell_type": "markdown",
    340    "metadata": {},
    341    "source": [
    342     "We will use window_size of 24 months and obtain the desired inputs for our RNN model, as follows:"
    343    ]
    344   },
    345   {
    346    "cell_type": "code",
    347    "execution_count": 8,
    348    "metadata": {},
    349    "outputs": [],
    350    "source": [
    351     "window_size = 24"
    352    ]
    353   },
    354   {
    355    "cell_type": "code",
    356    "execution_count": 36,
    357    "metadata": {},
    358    "outputs": [],
    359    "source": [
    360     "X, y = create_multivariate_rnn_data(df_transformed, window_size=window_size)"
    361    ]
    362   },
    363   {
    364    "cell_type": "code",
    365    "execution_count": 37,
    366    "metadata": {},
    367    "outputs": [
    368     {
    369      "data": {
    370       "text/plain": [
    371        "((420, 24, 2), (420, 2))"
    372       ]
    373      },
    374      "execution_count": 37,
    375      "metadata": {},
    376      "output_type": "execute_result"
    377     }
    378    ],
    379    "source": [
    380     "X.shape, y.shape"
    381    ]
    382   },
    383   {
    384    "cell_type": "markdown",
    385    "metadata": {},
    386    "source": [
    387     "Finally, we split our data into a train and a test set, using the last 24 months to test the out-of-sample performance, as shown here:"
    388    ]
    389   },
    390   {
    391    "cell_type": "code",
    392    "execution_count": 38,
    393    "metadata": {},
    394    "outputs": [],
    395    "source": [
    396     "test_size =24\n",
    397     "train_size = X.shape[0]-test_size"
    398    ]
    399   },
    400   {
    401    "cell_type": "code",
    402    "execution_count": 39,
    403    "metadata": {},
    404    "outputs": [],
    405    "source": [
    406     "X_train, y_train = X[:train_size], y[:train_size]\n",
    407     "X_test, y_test = X[train_size:], y[train_size:]"
    408    ]
    409   },
    410   {
    411    "cell_type": "code",
    412    "execution_count": 40,
    413    "metadata": {},
    414    "outputs": [
    415     {
    416      "data": {
    417       "text/plain": [
    418        "((396, 24, 2), (24, 24, 2))"
    419       ]
    420      },
    421      "execution_count": 40,
    422      "metadata": {},
    423      "output_type": "execute_result"
    424     }
    425    ],
    426    "source": [
    427     "X_train.shape, X_test.shape"
    428    ]
    429   },
    430   {
    431    "cell_type": "markdown",
    432    "metadata": {},
    433    "source": [
    434     "## Define Model Architecture"
    435    ]
    436   },
    437   {
    438    "cell_type": "markdown",
    439    "metadata": {},
    440    "source": [
    441     "We use a similar architecture with two stacked LSTM layers with 12 and 6 units, respectively, followed by a fully-connected layer with 10 units. The output layer has two units, one for each time series. We compile them using mean absolute loss and the recommended RMSProp optimizer, as follows:"
    442    ]
    443   },
    444   {
    445    "cell_type": "code",
    446    "execution_count": 70,
    447    "metadata": {},
    448    "outputs": [],
    449    "source": [
    450     "n_features = output_size = 2"
    451    ]
    452   },
    453   {
    454    "cell_type": "code",
    455    "execution_count": 71,
    456    "metadata": {},
    457    "outputs": [],
    458    "source": [
    459     "lstm1_units = 12\n",
    460     "lstm2_units = 6"
    461    ]
    462   },
    463   {
    464    "cell_type": "code",
    465    "execution_count": 72,
    466    "metadata": {},
    467    "outputs": [],
    468    "source": [
    469     "rnn = Sequential([\n",
    470     "    LSTM(units=lstm1_units,\n",
    471     "         dropout=.2,\n",
    472     "         recurrent_dropout=.2,\n",
    473     "         input_shape=(window_size, n_features), name='LSTM1',\n",
    474     "         return_sequences=True),\n",
    475     "    LSTM(units=lstm2_units,\n",
    476     "         dropout=.2,\n",
    477     "         recurrent_dropout=.2,\n",
    478     "         name='LSTM2'),\n",
    479     "    Dense(10, name='FC1'),\n",
    480     "    Dense(output_size, name='Output')\n",
    481     "])"
    482    ]
    483   },
    484   {
    485    "cell_type": "markdown",
    486    "metadata": {},
    487    "source": [
    488     "The model has 1,268 parameters, as shown here:"
    489    ]
    490   },
    491   {
    492    "cell_type": "code",
    493    "execution_count": 73,
    494    "metadata": {},
    495    "outputs": [
    496     {
    497      "name": "stdout",
    498      "output_type": "stream",
    499      "text": [
    500       "_________________________________________________________________\n",
    501       "Layer (type)                 Output Shape              Param #   \n",
    502       "=================================================================\n",
    503       "LSTM1 (LSTM)                 (None, 24, 12)            720       \n",
    504       "_________________________________________________________________\n",
    505       "LSTM2 (LSTM)                 (None, 6)                 456       \n",
    506       "_________________________________________________________________\n",
    507       "FC1 (Dense)                  (None, 10)                70        \n",
    508       "_________________________________________________________________\n",
    509       "Output (Dense)               (None, 2)                 22        \n",
    510       "=================================================================\n",
    511       "Total params: 1,268\n",
    512       "Trainable params: 1,268\n",
    513       "Non-trainable params: 0\n",
    514       "_________________________________________________________________\n"
    515      ]
    516     }
    517    ],
    518    "source": [
    519     "rnn.summary()"
    520    ]
    521   },
    522   {
    523    "cell_type": "code",
    524    "execution_count": 45,
    525    "metadata": {},
    526    "outputs": [],
    527    "source": [
    528     "rnn.compile(loss='mae', optimizer='RMSProp')"
    529    ]
    530   },
    531   {
    532    "cell_type": "markdown",
    533    "metadata": {},
    534    "source": [
    535     "## Train the Model"
    536    ]
    537   },
    538   {
    539    "cell_type": "markdown",
    540    "metadata": {},
    541    "source": [
    542     "We train for 50 epochs with a batch_size value of 20 using early stopping:"
    543    ]
    544   },
    545   {
    546    "cell_type": "code",
    547    "execution_count": 46,
    548    "metadata": {},
    549    "outputs": [],
    550    "source": [
    551     "early_stopping = EarlyStopping(monitor='val_loss', \n",
    552     "                              patience=5,\n",
    553     "                              restore_best_weights=True)"
    554    ]
    555   },
    556   {
    557    "cell_type": "code",
    558    "execution_count": 47,
    559    "metadata": {},
    560    "outputs": [],
    561    "source": [
    562     "rnn_path = 'models/fred.lstm_{}_{}.weights.best.hdf5'.format(lstm1_units, lstm2_units)\n",
    563     "checkpointer = ModelCheckpoint(filepath=rnn_path,\n",
    564     "                              monitor='val_loss',\n",
    565     "                              save_best_only=True,\n",
    566     "                              save_weights_only=True,\n",
    567     "                              period=5)"
    568    ]
    569   },
    570   {
    571    "cell_type": "code",
    572    "execution_count": 50,
    573    "metadata": {},
    574    "outputs": [
    575     {
    576      "name": "stdout",
    577      "output_type": "stream",
    578      "text": [
    579       "Train on 396 samples, validate on 24 samples\n",
    580       "Epoch 1/50\n",
    581       "396/396 [==============================] - 2s 5ms/step - loss: 3.8474 - val_loss: 2.2458\n",
    582       "Epoch 2/50\n",
    583       "396/396 [==============================] - 1s 2ms/step - loss: 3.8011 - val_loss: 2.2199\n",
    584       "Epoch 3/50\n",
    585       "396/396 [==============================] - 1s 2ms/step - loss: 3.7493 - val_loss: 2.1867\n",
    586       "Epoch 4/50\n",
    587       "396/396 [==============================] - 1s 2ms/step - loss: 3.7158 - val_loss: 2.1477\n",
    588       "Epoch 5/50\n",
    589       "396/396 [==============================] - 1s 2ms/step - loss: 3.6758 - val_loss: 2.0966\n",
    590       "Epoch 6/50\n",
    591       "396/396 [==============================] - 1s 2ms/step - loss: 3.6266 - val_loss: 2.0551\n",
    592       "Epoch 7/50\n",
    593       "396/396 [==============================] - 1s 2ms/step - loss: 3.5552 - val_loss: 2.0113\n",
    594       "Epoch 8/50\n",
    595       "396/396 [==============================] - 1s 2ms/step - loss: 3.4837 - val_loss: 1.9549\n",
    596       "Epoch 9/50\n",
    597       "396/396 [==============================] - 1s 2ms/step - loss: 3.4383 - val_loss: 1.9126\n",
    598       "Epoch 10/50\n",
    599       "396/396 [==============================] - 1s 2ms/step - loss: 3.4017 - val_loss: 1.8605\n",
    600       "Epoch 11/50\n",
    601       "396/396 [==============================] - 1s 2ms/step - loss: 3.3652 - val_loss: 1.8217\n",
    602       "Epoch 12/50\n",
    603       "396/396 [==============================] - 1s 2ms/step - loss: 3.2948 - val_loss: 1.7720\n",
    604       "Epoch 13/50\n",
    605       "396/396 [==============================] - 1s 2ms/step - loss: 3.2134 - val_loss: 1.7650\n",
    606       "Epoch 14/50\n",
    607       "396/396 [==============================] - 1s 2ms/step - loss: 3.2003 - val_loss: 1.7255\n",
    608       "Epoch 15/50\n",
    609       "396/396 [==============================] - 1s 2ms/step - loss: 3.1277 - val_loss: 1.7235\n",
    610       "Epoch 16/50\n",
    611       "396/396 [==============================] - 1s 2ms/step - loss: 3.1009 - val_loss: 1.7312\n",
    612       "Epoch 17/50\n",
    613       "396/396 [==============================] - 1s 2ms/step - loss: 3.0671 - val_loss: 1.7212\n",
    614       "Epoch 18/50\n",
    615       "396/396 [==============================] - 1s 2ms/step - loss: 3.0568 - val_loss: 1.7204\n",
    616       "Epoch 19/50\n",
    617       "396/396 [==============================] - 1s 2ms/step - loss: 3.0291 - val_loss: 1.7195\n",
    618       "Epoch 20/50\n",
    619       "396/396 [==============================] - 1s 2ms/step - loss: 3.0069 - val_loss: 1.7127\n",
    620       "Epoch 21/50\n",
    621       "396/396 [==============================] - 1s 2ms/step - loss: 2.9270 - val_loss: 1.7272\n",
    622       "Epoch 22/50\n",
    623       "396/396 [==============================] - 1s 2ms/step - loss: 2.9920 - val_loss: 1.7219\n",
    624       "Epoch 23/50\n",
    625       "396/396 [==============================] - 1s 2ms/step - loss: 2.8941 - val_loss: 1.7354\n",
    626       "Epoch 24/50\n",
    627       "396/396 [==============================] - 1s 2ms/step - loss: 2.8532 - val_loss: 1.7585\n",
    628       "Epoch 25/50\n",
    629       "396/396 [==============================] - 1s 2ms/step - loss: 2.9323 - val_loss: 1.7604\n"
    630      ]
    631     }
    632    ],
    633    "source": [
    634     "result = rnn.fit(X_train,\n",
    635     "                 y_train,\n",
    636     "                 epochs=50,\n",
    637     "                 batch_size=20,\n",
    638     "                 validation_data=(X_test, y_test),\n",
    639     "                 callbacks=[checkpointer, early_stopping],\n",
    640     "                 verbose=1)"
    641    ]
    642   },
    643   {
    644    "cell_type": "markdown",
    645    "metadata": {},
    646    "source": [
    647     "## Evaluate the Results"
    648    ]
    649   },
    650   {
    651    "cell_type": "markdown",
    652    "metadata": {},
    653    "source": [
    654     "Training stops early after 25 epochs, yielding a test MAE of 1.71, which compares favorably to the test MAE for the VAR model of 1.91."
    655    ]
    656   },
    657   {
    658    "cell_type": "markdown",
    659    "metadata": {},
    660    "source": [
    661     "However, the two results are not fully comparable because the RNN model produces 24 one-step-ahead forecasts, whereas the VAR model uses its own predictions as input for its out-of-sample forecast. You may want to tweak the VAR setup to obtain comparable forecasts and compare their performance:"
    662    ]
    663   },
    664   {
    665    "cell_type": "code",
    666    "execution_count": 51,
    667    "metadata": {},
    668    "outputs": [
    669     {
    670      "data": {
    671       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvDW2N/gAAIABJREFUeJzt3Xl4VOX99/H3mZkkk2SyJ5OE7CuEBAgQFmUJOwhGqAgtoCiPXdxKlf6goL9W1EdxK+Ly1Kq1rVUqakUiCrKWsMm+wxACBEIgmZAFspB95vnjxAQkkgBJJpl8X9d1rklmTs5852b4zJn73Oc+itVqtSKEEMKuaGxdgBBCiJYn4S6EEHZIwl0IIeyQhLsQQtghCXchhLBDEu5CCGGHJNyFEMIOSbgLIYQdknAXQgg7pLPVE+/btw9nZ2dbPX27VVlZiZOTk63LaHekXa4nbdI4e2+XyspKEhMTm1zPZuGuKApxcXG2evp2y2QySbs0QtrletImjbP3djGZTM1aT7plhBDCDkm4CyGEHZJwF0IIO2SzPnchROdUXV1NdnY2FRUVrbb95vZLt2d6vZ7g4GAcHBxu6e8l3IUQbSo7Oxs3NzfCw8NRFKXFt19eXt7hR+JZrVYKCgrIzs4mIiLilrYh3TJCiDZVUVGBj49PqwS7vVAUBR8fn9v6diPhLoRocxLsTbvdNrJZuF+uqCUzv8xWTy+EEHbNZuFuLq1h+OubGLU4jZdXH2fv2SIsFrmcqxCi9fXu3dvWJbQ6mx1QjfByZGFKd9aZzPxty2n+mnYKX4MjI7v5M7q7P4NjfNE7aG1VnhBCdGg2C3cHrcJDgyJ4aFAEl8ur2ZSex7pjZlYdzuGzPefQO2gYEuPH6O7+jOxmxMdgv3NFCCFsw2q18uqrr7JlyxYUReHRRx9l/Pjx5OXl8dRTT1FaWkptbS0LFy6kd+/ePPPMMxw5cgRFUZg8eTIPPfSQrV/CT2oXQyE9nB2YmBjExMQgqmos7MwsYN0xM+uPmVl3zIyiQN9QL0Z392dMfAARvq62LlkI0QK+3JvN53vOteg2J/XyZ9rAyGatu3btWo4fP05qaipFRUXcd999JCUl8c033zB48GAeffRRamtrKS8vx2QyYTab+eabbwAoLi5u0bpbWrsI96s56tQ99iExfjx3TzxHLxSrQW8ys2j1cRatPk5CkDspPbtwd68uBHl27PGsQgjb2bt3LxMmTECr1eLr60u/fv04fPgwPXr04Omnn6ampoZRo0YRFxdHSEgI586d44UXXiA5OZnBgwfbuvwbanfhfjVFUUgI8iAhyIOnRsdy/lI5qw/nsPJQTn3Q9w3zIqVnION7BmJ009u6ZCHETZjcN5jJfYNbdJvl5eW3vY1+/frxySefkJaWxvz585k1axaTJk0iNTWVrVu3smzZMlavXs2iRYtaoOLW0aHGuQd5OvPLIZGkPj6IzXOHM3dsV8oqa1i48hgDX9rA9A928OmuLIrKqmxdqhCiA0hKSmL16tXU1tZSWFjInj176NmzJ+fPn8fX15epU6cyZcoUjh49SmFhIVarlbFjx/Lkk09y7NgxW5d/Q+16z/1GQn1ceHx4NI8PjybDXMLKQzmsPHiBBcsP88cVRxga60dKr0BGxfnjpr+1uRmEEPZt9OjR7N+/n4kTJ6IoCnPnzsXPz4+vvvqKDz/8EJ1Oh4uLC6+88gp5eXksWLAAi8UCwJw5c2xc/Y0pVqvVJoPL9+/f3+JjTa1WK0cvFLPy0AW+OZjD+UvlOOk0jOhmZEy8P31DvQnxdm7XZ8fZ+4UGbpW0y/U6apu0dt32MLfMDxprq+a2X4fdc2/M1X30fxjbjf3nLrHy4AW+PZzD6iO5APganOgT6kmfMC/6hHrRM9hDxtMLIeyOXYX71TQahb5hXvQN8+KPd3cnPbeEfVlF6nK2iLXHzADoNArdu7jTJ9SL3qGe9An1Itirfe/dCyFEU+w23K+mrQvw7l3cuX9gGAAFpZXsz7pUH/if7T7HP7efAcDPTd277xvmRVK4Nz2DPNBpO9SxZyFEJ9cpwr0xPgYnRnX3Z1R3fwBqai0czy1hf1YRe88WsS/rEmuOqnv3bk46Bkb5MDjal0HRvkT5ucqevRCiXeu04f5jOq2mvr/+gTvCAbhYUsmuzEK2nsxn28l81tV15QS46xkU7cvgGB8GRflidJfx9UKI9kXC/Qb83JyY0DOQCT0DAcgquMK2U/lsPZnPxuNmvtyXDUCM0aCGfbQvAyK9ZeilEMLmJNxvQqiPC6E+oUzrH4rFYuVYTjHbTuaz7VQBy3Zn8c/tZ9BqFBJDPBkW68ddPQKINrrZumwhRCck4X6LNJqGYZe/SY6isqaWfWcvse1kPltO5rN4/Qn+vO4EMUYDdyUEcFePQLoFuElfvRAdTO/evdm/f3+jj2VnZ/PII4/UTybWnki4txAnnZY7ony4I8qH/xnbFXNxBWuO5rLqcA7v/Pckb208SbiPC3f1COSuhAB6BHlI0AshWo2Eeyvxd9cz845wZt4RTn5pJWuPmll9JIf3N5/m3U2nCPJ0rt+j7x3iiUYjQS86oQOfwv5PWnST2oSp0O/Bn3z89ddfJzAwkBkzZgDw9ttvo9Vq2blzJ8XFxdTU1PC73/2OUaNG3dTzVlZWsnDhQo4cOYJWq2X+/PkMHDiQjIwMFixYQHV1NRaLhbfffhuj0ciTTz5Jbm4uFouFxx57jPHjx9/W6/4xCfc24GtwYvqAUKYPCKWorIp1JjPfHcnlo+/P8LetmQS46xmXEMC4hAAMcqlBIVrV+PHjeemll+rDffXq1Xz44YfMnDkTg8FAYWEhP//5zxk5cuRNfbteunQpACtXruTUqVM8/PDDrFmzhmXLljFz5kzuueceqqqqsFgspKWlYTQaef/99wEoKSlp8dcp4d7GvFwdmZoUwtSkEIorqtlgMrP6cC7/3qUekA0w6HiixJX7+gbLtAjC/iVOU5cWVNvElL/du3enoKAAs9lMUVER7u7u+Pr6smjRInbv3o1Go8FsNpOfn4+fn1+zn3fv3r3cf//9AERFRdGlSxcyMzNJTEzkr3/9K7m5uYwZM4bw8HBiY2N55ZVXeO211xg+fDhJSUm39ZobI6dd2pC73oGf9Q7m/ZlJ7P/jaN78RSIeei3/u+IIQ179L++lnaK0ssbWZQphd8aNG8eaNWtYtWoV48ePZ+XKlRQWFrJ8+XJSU1Px9fWlsrKyRZ4rJSWFd999F71ez69//Wu+//57IiIiWL58ObGxsSxZsoR33nmnRZ7rahLu7YSrk46JiUG8Mb4LS385gFh/A4tWH+fORRtYvDadQpmjXogWM378eFatWsWaNWsYN24cJSUl+Pj44ODgwI4dOzh//vxNbzMpKYmVK1cCkJmZSU5ODpGRkZw7d46QkBBmzpzJyJEjSU9Px2w24+zszMSJE3n44YdbZW546ZZpZxRFYVDdNAcHzl3iL3UjbT7Ykskv+ofwqyGRdJFLCwpxW2JiYigrK8NoNGI0GklJSeHRRx8lJSWFhIQEIiObdw3Wq02fPp2FCxeSkpKCVqtl0aJFODo6snr1alJTU9HpdPj6+vKb3/yGw4cP8+qrr6LRaNDpdCxcuLDFX6NdzeduDxqbqznDXMK7aadIPXABjQI/6x3EI8lRRPoZbFRl2+uoc5e3po7aJjKfe/Pdznzu0i3TAcT4u7F4aiKb/mcY0/qHknrgAiMXp/HY0r0cOX/Z1uUJIdoh6ZbpQEK8XXh+YgK/HRHDP7Zl8vH3Z1l1OJehsX48NiyKARHecmKUEK0gPT2defPmXXOfo6MjX3zxhY0qapqEewfk5+bEvHHdeGRYFJ/sOMvft2byi/d30DfMi8eGRTGim1FCXrRrVqu1Q71Hu3btSmpqaps+5+32mEu3TAfmrnfgsWHRbP3DCJ6fGE/u5Qoe/mgPd725hdQD56mptdi6RCGuo9frKSgouO3wsmdWq5WCggL0+lufTrzJPffKykpmzJhBVVUVtbW1jB07ltmzZ1+zzvLly3n11Vfx91cvfHH//fczZcqUWy5K3By9g5aZd4QzrX8oXx+4wLtpp/jdsgMsXneC3wyNYnLfIJx0ckKUaB+Cg4PJzs7m4sWLrbL96upqHBw6/rTber2e4ODgW/77JsPd0dGRjz76CFdXV6qrq5k+fTpDhw4lMTHxmvXGjx/Pn/70p1suRNw+B62GyX2D+VnvINYeM/OXTSd5+qvDvLnhBL8cHMn0AaG4OklPnLAtBwcHIiIiWm37HXUUUUtrsltGURRcXV0BqKmpoaampkP1lXVGGo3CuIQAUh8fxCcPDyDS18CLq0wMemUjb6w7QZGcECWE3WvWOPfa2lruvfdesrKymD59OnPnzr3m8eXLl7N48WK8vLyIiIhgwYIFBAYG3nCbe/fuxcXF5faqt0MVFRW31c/2U0wXK/j88CV2nLuCXqcwPtade+M98HHpGHvyrdUuHZm0SeM6Q7s055vJTZ3EVFxczOOPP84f//hHYmNj6+8vKirC1dUVR0dHli1bxqpVq/jXv/51w23JSUyNa+2vlOm5Jby76SQrD+WgVRRGdTcyKs6f4V2NeLk6ttrz3i75qn09aZPG2Xu7NPf13dRum7u7OwMGDGDLli3XhLuXl1f9z1OmTOG11167mc2KNtQ1wI0lv+jNnNFd+XDraVYdyWXV4Vw0CvQN82JknD+j4oxE+Rmk+02IDqzJcC8sLESn0+Hu7k5FRQXbt2/nV7/61TXr5OXlYTQaAdi4cSNRUVGtU61oMaE+Ljw3MYFnU+I5fP4yG0xm1pvyeHn1cV5efZwwHxdGxfkzMs5Iv3BvHLQyalaIjqTJcM/Ly2P+/PnU1tZitVoZN24cw4cP58033yQhIYGRI0fy8ccfs3HjRrRaLR4eHixatKgtahctQKNR6BXiSa8QT+aM6cqFS+VsOJ7HBpOZj78/y4dbM3HX60juamRUnJFhsUY8XDr+MDMh7J1MHNbOtKf+wrLKGrZk5LPBZOa/6Xnkl1ah1Sj0C/diSIwfAyK86RHs0SZj6NtTu7QX0iaNs/d2aZU+d9G5uDrp6i//Z7FYOZB9iQ0mMxtMeby2Jh0AJ52G3qGeDIjwYUCEN71DvXB2lBOmhLA1CXfRLBqNQp9QL/qEejF3bDcKy6rYfaaQXZnq8vbGDN60gk6j0DPYg/51Yd833At3vXTjCNHWJNzFLfF2dWRsfABj4wMAKKmoZs/Zovqw/3Draf6adgqNAt27uNM/3If+Ed70j/DGux0PuRTCXki4ixbhpndgeFcjw7uqo6bKq2rZf66InafVsF+68yx/35YJQIzRUB/0AyJ8CPCw7xNOhLAFCXfRKpwdtdwZ5cudUb4AVNVYOJR9iV11XTmpBy6wdGcWAKHeLleFvTeh3i4yxl6I2yThLtqEo05DUrg3SeHePDYMamotHM8tYWdmIbsyC9hgMvOfvdkA+Ls70T/Cpz7sozvR5QSFaCkS7sImdFoNCUEeJAR58PDgCCwWK6culrIj84eDtAWsPHgBUPv3h4bqmRdYLhcHF6KZJNxFu6DRKMT4uxHj78YDA8OwWq1kFV5hZ2Yhm09c5OvDOXyT/l8mJgbxSHIkMf5uti5ZiHZNwl20S4qiEObjSpiPK1OTQti0+xCbcrQs253Fl/uyGd3dn0eSo+gb5tX0xoTohCTcRYfgb3Bg4T1xzB4Zw0fbz/DR92dYd8xM/whvHk2OYlhXPzkIK8RVZDYo0aF4uzry1OhYtv1hBH+6uzvnCq8w65+75bqxQvyIhLvokFyddPyfwRGkzR3O61N6UWOx8rtlBxj2+ib+9f0ZyqtqbV2iEDYl4S46NEedhvv6BrP2yaF8MDMJo5sTf0o9yqBXNvL2hgyKK6ptXaIQNiF97sIuaDQKo7urFxrZfaaIdzed5M/rTvD3bZk8MSKG+weGtsnslUK0F7LnLuyKoij0j/DmH7P6881vBxPfxYMXvjnGyD+nsWL/eSwWm8xwLUSbk3AXdishyINPfjmAjx/uj4ezA09+doAJb28l7cRFbHQZAyHajIS7sHtDYvxY+cRg3vxFIiUV1Tz4913M+NtODmdftnVpQrQaCXfRKWg0ChMTg9jw+2SeTenO8dwSUt7ZyhP/3sfZgjJblydEi5NwF52Kk07LrEERpM0dxm9HRLPBlMfIP6fxbOoR8ksrbV2eEC1Gwl10Sm56B34/pitpc4cxtV8In+zMIvnV//Lm+gzKKmtsXZ4Qt02GQopOzeiu56Wf9eDhwRG89l06b6w/wb++P0P3Lu74GZzwMTjiY3DCx9URX4MTvnX3ebs6oneQoZWi/ZJwFwKI8jPw1wf6si+riA+3ZJJ9qZzM/DLySyupqG58SgM3Jx0+Bsf6wPd31/Oz3kH0DpXJzITtSbgLcZU+oV70mdEQzlarlStVtRSUVpFfVqnellZSUFpJfmkVBWVV5JdUkplfxpaMfP71/VkGR/vyxIhoBkR4y2RmwmYk3IW4AUVRcHXS4eqkI9TH5YbrllbWsHTHWT7Ykskv3t9BUpgXj4+IZliszFgp2p4cUBWihRicdPwmOYqtfxjOc/fEc+FSObP+sZuUd7by3ZEcOTtWtCkJdyFamN5By4N3hrNp7nBemdyD0ooaHvlkH2OXbGbFfpmWWLQNCXchWomjTsPP+4Wyfk4yb/4iEUWBJz87wIg/p/Hpriwqa2RaYtF6JNyFaGU6rYaJiUF897uhvPdAXzycHViw/DDDXtvEP7ZlytzzolXIAVUh2ohGozA2PoAx3f3ZnJHP/9t4kudWHuOtDRn0CfWiW6Ab3QLc6RbgRoSvKzqt7HuJWyfhLkQbUxSF5Fg/kmP92Hm6gE93ZXEsp5hNJy5SW3fQ1VGnIcZooGuAG3EB7nQLdKNrgBt+BicZeSOaRcJdCBsaEOnDgEgfACprajmZV0p6bgnHc0sw5RSzNSOf5fvO16/v4+qoBr2/O56U4uBTQqSvAY1GAl9cS8JdiHbCSaclvosH8V08rrm/oLSyPvCP5xZzPLeEf+86S0W1hcXbLuLh7EDvUE/6hnrRJ8yLXiGeGJzkv3ZnJ+8AIdo5H4MTd0Y7cWe0b/19tRYr63ce4rLOm31ZRezLKmJT+kUANAp0DXCnT6gnfcO86BPqRZiPi3TndDIS7kJ0QFqNQqinI3FxIUztFwLA5fJqDpy7xL6zath/feACS3dmAWp3Tu9QL/qEedIr2BNfgxMezg54ujjIBGh2SsJdCDvh4exQf6AW1L37k3ml7K0L+31ZRaw3ma/7O0edBk9nh/qw93B2wMPZ8Ue/O+BrcCI2wIDRTd/WL03cAgl3IeyUVqPQNUAdZTN9QCgARWVVmHKKKbpSzeXyai6VV3G5vJrLP/x+pZoLlyow5ZRwubya0kbmtvc1OBIX6F63uBEX6E6UnwEHGbrZrki4C9GJeLk6XtN335TqWgvF5dVcKq8mr7iS47nFHLtQjCm3mH9uP0NVjTqVgqNWQ7TRQPcuDaHfPdAdTxfH1nopoglNhntlZSUzZsygqqqK2tpaxo4dy+zZs69Zp6qqinnz5nH06FE8PT154403CA4ObrWihRBtw0GrUS9WYnAiys/AHVE+9Y/V1Fo4nV+GKaeYYznFmHJK2JR+kf/sza5fJ9BDT9cANyJ9DUT4uRLp60q4ryuB7noZvtnKmgx3R0dHPvroI1xdXamurmb69OkMHTqUxMTE+nW++OIL3N3dWbduHd9++y2vv/46S5YsadXChRC2pdNqiPV3I9bfjYmJQfX3XyypxJRTXL+km0vZebqQ8uqGaRacdBoifF2JqAv7CF81+CN8XfF2dZSRPS2gyXBXFAVXV1cAampqqKmpua7hN27cyBNPPAHA2LFjef7557FarfIPJEQn5OfmhJ+bH0PrDuyCetETc3Elp/NLycwvI/NiGWcKykg3l7DumJmaq6ZDdtPriPR1JSHIg98MjWpyHn3RuGb1udfW1nLvvfeSlZXF9OnT6dWr1zWPm81mAgMD1Q3qdLi5uVFUVIS3t3fLVyyE6HAURSHAQ0+Ah547o67t86+ptXD+Ujmn60I/M18N/v/szebzPeeY1j+UJ0ZEt+oonbziCiqqLXb1QdKscNdqtaSmplJcXMzjjz/OiRMniI2Nva0ntlgsmEym29qGPaqoqJB2aYS0y/XsrU0CgAAfuMNHC7iT38uFTw8V8cmOs3y2O4tJcR7cl+CBwfHG4/Kb2y5Wq5Xj+ZWkHrvM1rNlWKxwT5w7D/X2Ru/Q8Uf+3NRoGXd3dwYMGMCWLVuuCXd/f39ycnIICAigpqaGkpISvLxufJFgjUZDXFzcrVVtx0wmk7RLI6RdrtcZ2mRIEmTml7F43Qk+O3iB706W8diwKB68M/wnT75qql2qaix8e/gC/9x2hoPZl3HT65g1KILKGgsf7zjL3txqXr63J4Njmj+qqC019wO9yY+nwsJCiouLAfUTcfv27URGRl6zzogRI/jqq68AWLNmDQMHDpT+diFEi4jwdeXtab355reD6R3qyaLVx0l+7b/8e2cW1TdxVau8kgqWrD/BoFc28tRnBymprOGFifHsWDCS/727Oy9MSuDz39yBg1bD/R/uZN5/DnK5vLoVX1nranLPPS8vj/nz51NbW4vVamXcuHEMHz6cN998k4SEBEaOHMl9993H3LlzGT16NB4eHrzxxhttUbsQohNJCPLgn7P6s/N0Aa+uSefprw7z/uZT/H5MVyb0CPzJoZWHsi/xz21nWHnoAtW1VoZ39eOhQREMifa97m/6R3iz+ndDeHNDBu9vPs2m9Iu8MCmBsfEBbfESW5RitVptctXe/fv307t3b1s8dbvWGb5q3wppl+t15jaxWq1sMOXx2pp00s0lxHdxZ+7YriTH+nH8+HGiY7vy3ZFc/rn9DHvPFuHqqGVKUggz7wgj0s/QrOc4nH2ZeV8ewpRTzIQegSy8Jx4/N6dWfmVNa+6/u5yhKoTocBRFYVR3f4Z3M/L1wfP8ee0JHvrHbgZEeBPlbmHjV/8lt7iCcB8Xnk3pzn19g3HTO9zUc/QI9uDrJwbxXtop3tpwkm2n8nk2pTuTEoNs0u184VI5728+zc9jmnewV8JdCNFhaTUKP+sdzIQeXVi2O4u3NpxkZ2YlQ2J8eeneBIbFGm/rTFgHrYYnRsQwLiGAef85xFOfHST1wAVe/FkPgjydW/CV/LSC0kr+sukUH+84C1b4eUx4s/5Owl0I0eE56jTMvCOcKX1D2HPoGEOSerTo9qONbnzxyJ386/szvPpdOmMWpzF/fBwz+oe22jQKJRXVfLAlkw+3nKa8upb7+gYze2QMJblnm/X3Eu5CCLvh7KjF17V1Yk2rUZg1KIJRcf4sWH6YP644wsoDF3huYjzdAtxarKumorqWf31/hr9sOsWlK9VM6BHIU6NjiTaqxwpMuc3bjoS7EELchBBvFz5+uD9f7MnmhW+PcdebWwjydGZIjC9DYvwYFO1zS7NhVtda+HzPOd7akIG5uJLkWD/+Z0xXegR7NP3HjZBwF0KIm6QoClP7hTC8m5Hvjuay5cRFvj2Uw7Ld51AU6BnkweC6sO8T6oWj7qcPglosVr4+eIHF606QVXiFpDAv3vpF7/oLp98qCXchhLhFfm5OPDAwjAcGhlFTa+Fg9iW2ZOSzNSOfv6ad5v/99xQujloGRvowONqXobG+RPkZUBQFq9XKelMef16bzvHcEuIC3fnHQ/0Y1tWvRbp4JNyFEKIF6LQa+oZ50zfMmydHxVJcUc2OUwVsPZnPlox8Nh7PAyDAXc/gGF9OXSxlf9Ylwn1ceGtab+6+wYlYt1RPi21JCCFEPXe9A2PiAxhTd3brucIrbD2p7tWvN5lxdtCy6N4e3Nc3uFUuUSjhLoQQbSDE24Vp/UOZ1j8Ui8WKotCqJ0NJuAshRBtri0sMdvxJi4UQQlxHwl0IIeyQhLsQQtghCXchhLBDEu5CCGGHJNyFEMIOSbgLIYQdknAXQgg7JOEuhBB2SMJdCCHskIS7EELYIQl3IYSwQxLuQghhhyTchRDCDkm4CyGEHZJwF0IIOyThLoQQdkjCXQgh7JCEuxBC2CEJdyGEsEMS7kIIYYck3IUQwg5JuAshhB2ScBdCCDsk4S6EEHZIwl0IIeyQrqkVcnJymDdvHgUFBSiKwtSpU3nwwQevWWfnzp089thjBAcHAzB69GieeOKJ1qlYCCFEk5oMd61Wy/z584mPj6e0tJTJkyczaNAgoqOjr1kvKSmJ9957r9UKFUII0XxNdssYjUbi4+MBMBgMREZGYjabW70wIYQQt+6m+tyzs7MxmUz06tXruscOHDjAPffcwy9/+UsyMjJarEAhhBA3T7FardbmrFhWVsYDDzzAI488wpgxY655rLS0FEVRcHV1JS0tjRdffJG1a9fecHt79+7FxcXl1iu3UxUVFej1eluX0e5Iu1xP2qRxnaFd4uLimlynyT53gOrqambPnk1KSsp1wQ5qd80PkpOTee655ygsLMTb2/snt6nRaJpVYGdjMpmkXRoh7XI9aZPG2Xu7mEymZq3XZLeM1WrlmWeeITIyklmzZjW6zsWLF/nhC8ChQ4ewWCx4eXndRLlCCCFaUpN77nv37iU1NZXY2FgmTpwIwJw5c7hw4QIA06ZNY82aNXz66adotVr0ej2LFy9GUZTWrVwIIcRPajLck5KSSE9Pv+E6999/P/fff3+LFSWEEOL2yBmqQghhhyTchRDCDkm4CyGEHZJwF0IIOyThLoQQdkjCXQgh7JCEuxBC2CEJdyGEsEMS7kIIYYck3IUQwg5JuAshhB2ScBdCCDsk4S6EEHZIwl0IIeyQhLsQQtghCXchhLBDEu5CCGGHbBbuSm0V1NbY6umFEMKuNXmZvdbidPkUvHQ3+HYFY1zd0l299QgBjXypEEKIW2WzcK9y7QIDH4U8E5zdDoc/b3jQ0QB+3dSg949vCH5XP5ALbwshRJNsFu4WJw8Y+HzDHRWXIe845B1TAz/vGKSvgv0fN6zj4qOGfXA/CO4PwUng6tv2xQshRDtns3C/jt4DQgeoy9WgZR/pAAAPeElEQVRKL14b+DkHYOsSsNaqj3tFQEj/usBPAv8E0Dq0ff1CCNGOtJ9w/ykGPzAkQ2Ryw31VV9SQP7cLsnfD6U1w6DP1MZ0zdOmtBn1wPzX43QJsUroQQthK+w/3xji6QNid6gJgtcLlc2rQZ+9RQ3/Hu2B5S33cIwRCBkDEUHXxCpe+eyGEXeuY4f5jigKeoeqSMFm9r7oCcg/XBf4uOLMFjvxHfcwjFCKHQkQyhA8B90Db1S6EEK3APsK9MQ56COmnLjym7t3nn4DMzWo3jukb2P+Juq5vrBr0EUMhfDC4eNuyciGEuG32G+4/pijg11Vd+v8KLLXqnn1mmhr4B5bC7g8ABQJ6qH38Eclq14+jq62rF0KIm9J5wv3HNFrokqgug34HNVVwYR+crgv7ne/B9rfVA7SxYyD+XogZo/b3CyFEO9d5w/3HdI4QOlBdhv1BHZFzbgcc/xaOpaqLgyt0HacGffQotetHCCHaIQn3n+LoAlEj1GXcK3B2GxxdDse+hiNfgpM7dB0PCfdC5HD1w0EIIdoJCffm0OrUPvjIZBj/utpPf+QrOL4SDi0DvSfE3a3u0Uckq+sLIYQNSQrdLK2D2iUTPQpq3oBTG+HoV3A0VR194+IDcfdAXIp6MNbB2dYVCyE6IQn326FzVPvgu45Tx9WfXK923Rz6HPb+A3R6NeCjRqrdO8Y4OXlKCNEmJNxbioNe7ZqJu1s9GHt2O5zaACc3wNpn1HXcAhv68SOHg6uPbWsWQtgtCffW4OgCMaPUBeByttp9c2qjOvrmwFJAgcBeEF23Vx/cXw7KCiFajIR7W/AIhj4z1cVSCxcOqHv1pzaqM1xu+bM6h334ELwM8RAwE7zCbF21EKIDk3BvaxotBPdVl+R56jz2mVvqu3ACTqyGfa+rFyuJGQ0xY9Wx9zKNsRDiJki425reo6Gv3mrl1K41RFlOwYk1sOOv6lmyTu4QNVw9QzZ6NLj527pqIUQ712S45+TkMG/ePAoKClAUhalTp/Lggw9es47VauXFF18kLS0NvV7Pyy+/THx8fKsVbbcUhSr3MIgbB3c8DpUl6iRnGWshY516lixAYCLEjlXDvksfud6sEOI6TYa7Vqtl/vz5xMfHU1payuTJkxk0aBDR0dH162zevJkzZ86wdu1aDh48yMKFC/niiy9atfBOwclNHS8fl6LOapl7GDLWqEG/+TVIewVcfNUx93EpatjLQVkhBM0Id6PRiNFoBMBgMBAZGYnZbL4m3Dds2MCkSZNQFIXExESKi4vJy8ur/zvRAhQFAnuqy9C5cKVQHWaZsVZdDi1TT6DqMRUSp6vrCSE6rZvqc8/OzsZkMtGrV69r7jebzQQENFzKLiAgALPZfMNwt1gsmEymmyzX/lVUVDS/XRwSoHsCdJuNIXcnHpnfYtj9NzQ736XCM4ZL4RMoDhtDrb7jz09/U+3SSUibNE7aRdXscC8rK2P27Nk8/fTTGAyG235ijUZDXFzcbW/H3phMpltrl/gewC/VPfojX6I/8G8CDiwh4NA7andN4nR15E0H7ba55XaxY9ImjbP3dmnuB1ezwr26uprZs2eTkpLCmDFjrnvc39+f3Nzc+t9zc3Px95cRHTbh4q1ejKT/ryDvOBz8Nxz8DNJXgbM39JwKvaapJ1DJVAhC2K0mh1lYrVaeeeYZIiMjmTVrVqPrjBgxghUrVmC1Wjlw4ABubm7S394eGLvB6OfhqaMw4z/qrJZ7/gHvJ8O7g2D7O+rZs0IIu9PknvvevXtJTU0lNjaWiRMnAjBnzhwuXLgAwLRp00hOTiYtLY3Ro0fj7OzMSy+91LpVi5uj1dWdEDUayovgyHI48G91zpu1z4BfnDoNQvRICL1TLkIihB1oMtyTkpJIT0+/4TqKovDss8+2WFGiFTl7Qb+H1SU/Qz1Z6uR62PU+fP+OelnBiCEN0xp7R0r3jRAdkJyh2pn5xqjLnU9AVRmc2aYG/cn16vBKAK/whqAPHwJOt38wXQjR+iTchcrRVb0QeGzdAfPC0+o4+pMb4MCnsPtvoHGAsDvUoO86AXyjb7xNIYTNSLiLxnlHQv9IddRNTSVk7VD36E9thHV/UpfAXpBwn3odWY9gW1cshLiKhLtoms6p4RqyvACXz8OxFeqFwtf9UV1C74CEydB9Ehj8bF2xEJ2ehLu4eR5B6sRmdzyudt8c+VIdgbPqf2D1PPUi4T3ug253g7OnrasVolOS6QTF7fGOVOe6eex7ePR7GDwHijIh9XF4PQY+nQ6H/6MesBVCtBnZcxctx7+7uoz4X7iwDw5/qV4wPP1bcHCBrndB7F3qUEu3gKa3J4S4ZRLuouUpCgT1VZcx/xeytqtdN8dS1VsA31iIGKou4UPUaROEEC1Gwl20Lo0Gwgery/jX1TnpMzeryw9DLFEgIEHtqw8fAmF3gt7d1pUL0aFJuIu2o9FCl0R1GTQbaqvh/D416M9shl0fqGfJKlro0rthzz5kgK0rF6LDkXAXtqN1gNAB6pI8F6orIHtXw5799rdg62LQOhJi7AsV90PX8dKFI0QzSLiL9sNB37C3Duo1ZLN2wOlNOB76Uh2Bo9Gp3Tfxk9ShlhL0QjRKwl20X05u9bNZngqZQZxHhXry1NEV8PVvYeWT6gdB94nqNWRdfW1dsRDthoS76BgUBYL6qMuo5yDnYEPQf/MkfPt79aBt94kQd4+cJSs6PQl30fEoSsOB2ZHPqiNwjqWqYf/tHPVM2bBBash7R6iTojm6gqOh4WcHF/UArxB2SsJddGyKAoE91WXE/0LeMXVv/tgKWD33xn+rc248+J3c1G8IEcnq5GjyISBaWk0VVFwGay1YasBSW/ezpe629ke3V91P844zSbgL+6Eo4B+vLiOegcJMuFIAVaXq9AdVZVf9fOUn7i+DS1lg+lrdpt5DHXsfOUwNe98YuXiJaB6rFcryoSBDvTBO/gkoOKn+XHRGDetb8fMdzVpNwl3YL+8IdbkVpXnqcMzTmyAzDY5/o97vFqiGfGSyeusR1GLlig6qpkqdQK8+xDMafq641LCe1gl8otUT9uJ/pk7BoWjUb4YanXp+h0bbcF/97z+6v7J5ZUm4C9EYg1Gd2bLHfervhZlqyJ9OU+e1P7RMvd8nuiHswwars2Aqmlvfu7daobpcHQZaVXrVbel1v/uZz8PFMNB7qpdP1Huqz19/66GeS9BZ1FRBWR6Ol09DoZPa7eagV291Tjf3b2KxQHmh+iFflgelF+tu86DsYsNt2UUoyb12L9wtUH1fJNwLPjHqVBu+0eAR0jJdfCZTs1aTcBeiOX74FtD3IfU/ft6xhrA/9Bns+fDa9ZW6vbH6vbIf76Hp1KkZNDpAgeordQFeAlZL80rSOMDx6huv5GhQw17v0RD8ene1BgX1uRVFra+xn1EaPqyUuklkFc2P7lOuv+/qv9XowNEFHH44vuGi1uXgctUxj584yG21QmWxGqYluVBqVpeSXPW+0lwoqbuvvBCAKIDvftwQCuj0DWHf2K3V0hDiZfmNd5toHMDVTx2N5WpUuwDdu6gB7hOtLu1k6gwJdyFulkajfrUOSFDntP9hGoVzO9W9bktNw8EvS40aGj8cNKt/zNLws9WiBp+TQQ29+lu3q353U3//4TFHV9LTTxAXEwnll9Sv/z++rbh8/X1FZ9S9f6ul7kPEqgYo1rr7GvuZ69e/5ver1r1dOue68HcFFDXAa8qvX0/rCAZ/dfGJUi//aAgAg5HsghKC/X3Vf4uaikZuK9RtXn17pUB9Po8gdRSWwaiG9w8hbjCqoe7s1WGOuUi4C3G7rp5Goa3pnMDNX11szfrjD4q60LdUqwewq8uuPZhdfeWqA9pldb+X1j1epv69wdgQ4m7+9QF+o5AtMZkgLq5tX3s7JOEuhGgZinJV4F7dvaJXv3WINiVXYhJCCDsk4S6EEHZIwl0IIeyQhLsQQtghCXchhLBDEu5CCGGHJNyFEMIOSbgLIYQdUqxWawucM3zzDhw4gJOTky2eWgghOqzKykoSExObXM9m4S6EEKL1SLeMEELYIQl3IYSwQxLuQghhhyTchRDCDkm4CyGEHbLJfO6bN2/mxRdfxGKxMGXKFH7961/boox2Z8SIEbi6uqLRaNBqtSxfvtzWJdnEggUL2LRpEz4+PnzzjXph6kuXLvHUU09x/vx5goKCWLJkCR4eHjautO001iZvv/02n3/+Od7e3gDMmTOH5ORkW5bZ5nJycpg3bx4FBQUoisLUqVN58MEHO/37BQBrG6upqbGOHDnSmpWVZa2srLSmpKRYMzIy2rqMdmn48OHWgoICW5dhc7t27bIeOXLEOmHChPr7XnnlFet7771ntVqt1vfee8/66quv2qo8m2isTd566y3r3/72NxtWZXtms9l65MgRq9VqtZaUlFjHjBljzcjI6PTvF6vVam3zbplDhw4RFhZGSEgIjo6OTJgwgQ0bNrR1GaId69ev33V7WRs2bGDSpEkATJo0ifXr19uiNJtprE0EGI1G4uPjATAYDERGRmI2mzv9+wVs0OduNpsJCAio/93f3x+z2dzWZbRbDz/8MPfeey+fffaZrUtpVwoKCjAajQD4+flRUFBg44rah6VLl5KSksKCBQu4fPmyrcuxqezsbEwmE7169ZL3C3JAtV359NNP+eqrr/jggw9YunQpu3fvtnVJ7ZKiKCgd5Ar0rWnatGmsW7eO1NRUjEYjL7/8sq1LspmysjJmz57N008/jcFguOaxzvp+afNw9/f3Jzc3t/53s9mMv387uHJ7O/BDO/j4+DB69GgOHTpk44raDx8fH/Ly8gDIy8urP4jYmfn6+qLVatFoNEyZMoXDhw/buiSbqK6uZvbs2aSkpDBmzBhA3i9gg3Dv0aMHZ86c4dy5c1RVVfHtt98yYsSIti6j3bly5QqlpaX1P2/bto2YmBgbV9V+jBgxghUrVgCwYsUKRo4caeOKbO+H8AJYv359p3y/WK1WnnnmGSIjI5k1a1b9/fJ+sdHEYWlpabz00kvU1tYyefJkHn300bYuod05d+4cjz/+OAC1tbXcfffdnbZd5syZw65duygqKsLHx4ff/va3jBo1iieffJKcnBy6dOnCkiVL8PT0tHWpbaaxNtm1axfHjx8HICgoiOeff76+n7mz2LNnDzNmzCA2NhaNRt1XnTNnDj179uzU7xeQWSGFEMIuyQFVIYSwQxLuQghhhyTchRDCDkm4CyGEHZJwF0IIOyThLoQQdkjCXQgh7JCEuxBC2KH/D5p9oGzFztZKAAAAAElFTkSuQmCC\n",
    672       "text/plain": [
    673        "<Figure size 432x288 with 1 Axes>"
    674       ]
    675      },
    676      "metadata": {},
    677      "output_type": "display_data"
    678     }
    679    ],
    680    "source": [
    681     "pd.DataFrame(result.history).plot();"
    682    ]
    683   },
    684   {
    685    "cell_type": "code",
    686    "execution_count": 52,
    687    "metadata": {},
    688    "outputs": [
    689     {
    690      "name": "stdout",
    691      "output_type": "stream",
    692      "text": [
    693       "<class 'pandas.core.frame.DataFrame'>\n",
    694       "DatetimeIndex: 24 entries, 2016-01-01 to 2017-12-01\n",
    695       "Data columns (total 2 columns):\n",
    696       "ip           24 non-null float32\n",
    697       "sentiment    24 non-null float32\n",
    698       "dtypes: float32(2)\n",
    699       "memory usage: 384.0 bytes\n"
    700      ]
    701     }
    702    ],
    703    "source": [
    704     "y_pred = pd.DataFrame(rnn.predict(X_test), columns=y_test.columns, index=y_test.index)\n",
    705     "y_pred.info()"
    706    ]
    707   },
    708   {
    709    "cell_type": "code",
    710    "execution_count": 65,
    711    "metadata": {},
    712    "outputs": [],
    713    "source": [
    714     "test_mae = mean_absolute_error(y_pred, y_test)"
    715    ]
    716   },
    717   {
    718    "cell_type": "code",
    719    "execution_count": 69,
    720    "metadata": {
    721     "scrolled": false
    722    },
    723    "outputs": [
    724     {
    725      "data": {
    726       "image/png": "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\n",
    727       "text/plain": [
    728        "<Figure size 1152x360 with 3 Axes>"
    729       ]
    730      },
    731      "metadata": {
    732       "needs_background": "light"
    733      },
    734      "output_type": "display_data"
    735     }
    736    ],
    737    "source": [
    738     "fig, axes = plt.subplots(ncols=3, figsize=(16, 5))\n",
    739     "pd.DataFrame(result.history).plot(ax=axes[0], title='Train & Validiation Error')\n",
    740     "axes[0].set_xlabel('Epoch')\n",
    741     "axes[0].set_ylabel('MAE')\n",
    742     "for i, col in enumerate(y_test.columns, 1):\n",
    743     "    y_train.loc['2010':, col].plot(ax=axes[i], label='training', title=col)\n",
    744     "    y_test[col].plot(ax=axes[i], label='out-of-sample')\n",
    745     "    y_pred[col].plot(ax=axes[i], label='prediction')\n",
    746     "    axes[i].set_xlabel('')\n",
    747     "plt.legend()\n",
    748     "fig.suptitle('Multivariate RNN - Results | Test MAE = {:.2f}'.format(test_mae), fontsize=16)\n",
    749     "fig.tight_layout()\n",
    750     "fig.subplots_adjust(top=.85)\n",
    751     "fig.savefig('figures/multivariate_results', dpi=300);"
    752    ]
    753   },
    754   {
    755    "cell_type": "markdown",
    756    "metadata": {},
    757    "source": []
    758   }
    759  ],
    760  "metadata": {
    761   "kernelspec": {
    762    "display_name": "Python 3",
    763    "language": "python",
    764    "name": "python3"
    765   },
    766   "language_info": {
    767    "codemirror_mode": {
    768     "name": "ipython",
    769     "version": 3
    770    },
    771    "file_extension": ".py",
    772    "mimetype": "text/x-python",
    773    "name": "python",
    774    "nbconvert_exporter": "python",
    775    "pygments_lexer": "ipython3",
    776    "version": "3.6.8"
    777   },
    778   "toc": {
    779    "base_numbering": 1,
    780    "nav_menu": {},
    781    "number_sections": true,
    782    "sideBar": true,
    783    "skip_h1_title": true,
    784    "title_cell": "Table of Contents",
    785    "title_sidebar": "Contents",
    786    "toc_cell": false,
    787    "toc_position": {},
    788    "toc_section_display": true,
    789    "toc_window_display": false
    790   }
    791  },
    792  "nbformat": 4,
    793  "nbformat_minor": 2
    794 }