{"cells":[{"metadata":{},"cell_type":"markdown","source":"<h2>Overview</h2>\n\nThere are quite a few discussions about the best validation method in LANL competition. The main argument against KFold (shuffle) is that segments from the same earthquake in train and validation sets could leak information about the later. This doesn't happen in the test set, since earthquakes are totally different from the training data.\n\nTo check this argument, I am trying to predict which earthquake a segment came from. I am not sure if this is the correct approach, so let me know your ideas about this experiment."},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"import time\nimport random\nimport numpy as np\nimport pandas as pd\nfrom tqdm import tqdm_notebook\n# seaborn and matplot\nimport matplotlib.pyplot as plt\nimport seaborn as sns\n# scipy (feature engineering)\nfrom scipy.signal import hilbert\nfrom scipy.signal import hann\nfrom scipy.signal import convolve\nfrom scipy import stats\nimport lightgbm as lgb\nimport warnings\n# Configurations\nwarnings.simplefilter(action='ignore', category=UserWarning)\nRANDOM_SEED = 19\nnp.random.seed(RANDOM_SEED)\nsns.set()\n\ndef plot_multiclass(result):\n    # Plot multi_logloss and 1 - multi_error\n    num_rounds = len(result['multi_logloss-mean'])\n    fig, ax1 = plt.subplots(figsize=(10, 5))\n    fig.suptitle('logloss (blue) and accuracy (orange)', fontsize=14)\n    ax2 = ax1.twinx()\n    ax1.set_xlabel('boosting round')\n    ax1.set_ylabel('logloss')\n    ax2.set_ylabel('accuracy')\n    p1 = sns.lineplot(x=np.arange(num_rounds), y=result['multi_logloss-mean'],\n                      ax=ax1, color='blue')\n    multi_accuracy = [1 - v for v in result['multi_error-mean']]  # not sure if this is right\n    p2 = sns.lineplot(x=np.arange(num_rounds), y=multi_accuracy, ax=ax2, color='orange')","execution_count":1,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"data_type = {'acoustic_data': np.int16, 'time_to_failure': np.float64}\ntrain = pd.read_csv('../input/train.csv', dtype=data_type)","execution_count":2,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<h2>Features</h2>\n\nI'm using a feature set similar to lukyanenko's kernel. Segments that belongs to two quakes are removed, so we have 4194 - 16 = 4178 data points."},{"metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","trusted":true},"cell_type":"code","source":"def extract_segment_features(frame, index, x):\n    frame.loc[index, 'std'] = x.values.std()\n    frame.loc[index, 'mean'] = x.values.mean()\n    frame.loc[index, 'max'] = x.values.max()\n    frame.loc[index, 'min'] = x.values.min()\n    frame.loc[index, 'std_abs'] = x.abs().std()\n    frame.loc[index, 'max_abs'] = x.abs().max()\n    frame.loc[index, 'mean_abs_change'] = np.mean(np.abs(x.diff()))\n    frame.loc[index, 'std_abs_change'] = np.std(np.abs(x.diff()))\n    \n    frame.loc[index, 'mad'] = x.mad()\n    frame.loc[index, 'iqr'] = stats.iqr(x.values)\n    frame.loc[index, 'kurt'] = x.kurtosis()\n    frame.loc[index, 'skew'] = x.skew()\n    frame.loc[index, 'q05'] = np.quantile(x, 0.05)\n    frame.loc[index, 'q95'] = np.quantile(x, 0.95)\n    \n    for windows in [16, 64, 512, 4096]:\n        x_roll_mean = x.rolling(windows).mean().dropna().values\n        x_roll_std = x.rolling(windows).std().dropna().values\n        frame.loc[index, 'q05_roll_std_' + str(windows)] = np.quantile(x_roll_std, 0.05)   \n        frame.loc[index, 'q95_roll_std_' + str(windows)] = np.quantile(x_roll_std, 0.95)\n        frame.loc[index, 'mean_roll_std_' + str(windows)] = np.mean(x_roll_mean)\n        frame.loc[index, 'q95_roll_mean_' + str(windows)] = np.quantile(x_roll_mean, 0.95)\n    return frame\n\n\ndef make_train(train_data, size=150000, skip=150000):\n    num_segments = int(np.floor((train_data.shape[0] - size) / skip)) + 1\n    # We will be removing segments that belongs to two quakes\n    num_segments -= 16\n\n    X_train = pd.DataFrame(index=range(num_segments), dtype=np.float64)\n    y_train = pd.DataFrame(index=range(num_segments), columns=['quake_number'])\n    quake_count = 0\n    \n    for index in tqdm_notebook(range(num_segments)):\n        seg = train_data.iloc[index*skip:index*skip + size]\n        \n        if any(seg.time_to_failure.diff() > 5):\n            quake_count += 1\n            continue\n        \n        y_train.loc[index, 'quake_number'] = quake_count\n        y_train.loc[index, 'time_to_failure'] = seg.time_to_failure.values[-1]\n        X_train = extract_segment_features(X_train, index, seg.acoustic_data)\n    return X_train, y_train","execution_count":3,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_tr, y_tr = make_train(train)\nX_tr.head()","execution_count":4,"outputs":[{"output_type":"display_data","data":{"text/plain":"HBox(children=(IntProgress(value=0, max=4178), HTML(value='')))","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"6fa91e37438d44da9040683df36f555f"}},"metadata":{}},{"output_type":"stream","text":"\n","name":"stdout"},{"output_type":"execute_result","execution_count":4,"data":{"text/plain":"        std         ...          q95_roll_mean_4096\n0  5.101089         ...                    5.239502\n1  6.588802         ...                    5.004395\n2  6.967374         ...                    5.256836\n3  6.922282         ...                    5.237061\n4  7.301086         ...                    5.181885\n\n[5 rows x 30 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>std</th>\n      <th>mean</th>\n      <th>max</th>\n      <th>min</th>\n      <th>std_abs</th>\n      <th>max_abs</th>\n      <th>mean_abs_change</th>\n      <th>std_abs_change</th>\n      <th>mad</th>\n      <th>iqr</th>\n      <th>kurt</th>\n      <th>skew</th>\n      <th>q05</th>\n      <th>q95</th>\n      <th>q05_roll_std_16</th>\n      <th>q95_roll_std_16</th>\n      <th>mean_roll_std_16</th>\n      <th>q95_roll_mean_16</th>\n      <th>q05_roll_std_64</th>\n      <th>q95_roll_std_64</th>\n      <th>mean_roll_std_64</th>\n      <th>q95_roll_mean_64</th>\n      <th>q05_roll_std_512</th>\n      <th>q95_roll_std_512</th>\n      <th>mean_roll_std_512</th>\n      <th>q95_roll_mean_512</th>\n      <th>q05_roll_std_4096</th>\n      <th>q95_roll_std_4096</th>\n      <th>mean_roll_std_4096</th>\n      <th>q95_roll_mean_4096</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>5.101089</td>\n      <td>4.884113</td>\n      <td>104.0</td>\n      <td>-98.0</td>\n      <td>4.333325</td>\n      <td>104.0</td>\n      <td>2.613217</td>\n      <td>2.168535</td>\n      <td>3.263401</td>\n      <td>4.0</td>\n      <td>33.662481</td>\n      <td>-0.024061</td>\n      <td>-2.0</td>\n      <td>11.0</td>\n      <td>1.927866</td>\n      <td>8.182705</td>\n      <td>4.884078</td>\n      <td>6.8125</td>\n      <td>2.387083</td>\n      <td>8.232413</td>\n      <td>4.883941</td>\n      <td>5.703125</td>\n      <td>2.650157</td>\n      <td>8.408263</td>\n      <td>4.883694</td>\n      <td>5.375000</td>\n      <td>2.873412</td>\n      <td>6.930391</td>\n      <td>4.879138</td>\n      <td>5.239502</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>6.588802</td>\n      <td>4.725767</td>\n      <td>181.0</td>\n      <td>-154.0</td>\n      <td>5.732777</td>\n      <td>181.0</td>\n      <td>2.701525</td>\n      <td>2.492046</td>\n      <td>3.574302</td>\n      <td>5.0</td>\n      <td>98.758517</td>\n      <td>0.390561</td>\n      <td>-2.0</td>\n      <td>12.0</td>\n      <td>1.932184</td>\n      <td>9.387048</td>\n      <td>4.725753</td>\n      <td>6.7500</td>\n      <td>2.396260</td>\n      <td>9.885744</td>\n      <td>4.725693</td>\n      <td>5.531250</td>\n      <td>2.633934</td>\n      <td>10.339422</td>\n      <td>4.725015</td>\n      <td>5.134766</td>\n      <td>2.845732</td>\n      <td>18.035823</td>\n      <td>4.721736</td>\n      <td>5.004395</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>6.967374</td>\n      <td>4.906393</td>\n      <td>140.0</td>\n      <td>-106.0</td>\n      <td>5.895945</td>\n      <td>140.0</td>\n      <td>2.792605</td>\n      <td>2.510750</td>\n      <td>3.948411</td>\n      <td>5.0</td>\n      <td>33.555211</td>\n      <td>0.217391</td>\n      <td>-3.0</td>\n      <td>13.0</td>\n      <td>1.995829</td>\n      <td>11.888370</td>\n      <td>4.906140</td>\n      <td>7.3750</td>\n      <td>2.457930</td>\n      <td>13.241655</td>\n      <td>4.906062</td>\n      <td>5.828125</td>\n      <td>2.695336</td>\n      <td>14.876549</td>\n      <td>4.906024</td>\n      <td>5.408203</td>\n      <td>2.919841</td>\n      <td>12.658163</td>\n      <td>4.903929</td>\n      <td>5.256836</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>6.922282</td>\n      <td>4.902240</td>\n      <td>197.0</td>\n      <td>-199.0</td>\n      <td>6.061214</td>\n      <td>199.0</td>\n      <td>2.705618</td>\n      <td>2.535464</td>\n      <td>3.647117</td>\n      <td>5.0</td>\n      <td>116.548172</td>\n      <td>0.757278</td>\n      <td>-2.0</td>\n      <td>12.0</td>\n      <td>1.950852</td>\n      <td>9.717124</td>\n      <td>4.902191</td>\n      <td>7.1250</td>\n      <td>2.418308</td>\n      <td>10.460610</td>\n      <td>4.902000</td>\n      <td>5.765625</td>\n      <td>2.665094</td>\n      <td>11.110177</td>\n      <td>4.901506</td>\n      <td>5.363281</td>\n      <td>2.901803</td>\n      <td>11.566750</td>\n      <td>4.899277</td>\n      <td>5.237061</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>7.301086</td>\n      <td>4.908720</td>\n      <td>145.0</td>\n      <td>-126.0</td>\n      <td>6.329485</td>\n      <td>145.0</td>\n      <td>2.712478</td>\n      <td>2.459512</td>\n      <td>3.826052</td>\n      <td>5.0</td>\n      <td>52.977905</td>\n      <td>0.064531</td>\n      <td>-2.0</td>\n      <td>12.0</td>\n      <td>1.949359</td>\n      <td>10.132250</td>\n      <td>4.908832</td>\n      <td>7.3750</td>\n      <td>2.416862</td>\n      <td>12.201375</td>\n      <td>4.909046</td>\n      <td>5.828125</td>\n      <td>2.664852</td>\n      <td>13.787718</td>\n      <td>4.909721</td>\n      <td>5.363281</td>\n      <td>2.972784</td>\n      <td>16.564661</td>\n      <td>4.909528</td>\n      <td>5.181885</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"plt.figure(figsize=(10, 5))\nplt.title(\"Correlation heatmap for features\")\nax = sns.heatmap(X_tr.corr(), annot=False, linewidths=.3, cmap=\"YlGnBu\")\nplt.figure(figsize=(10, 5))\nplt.title(\"Count of segments for each earthquake (y_tr)\")\nax = sns.countplot(x=\"quake_number\", data=y_tr, palette='GnBu_d')","execution_count":5,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 720x360 with 2 Axes>","image/png":"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\n"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 720x360 with 1 Axes>","image/png":"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\n"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"Let's also remove the first and last group; so we have 15 groups."},{"metadata":{"trusted":true},"cell_type":"code","source":"keep_idx = y_tr[(y_tr.quake_number > 0) & (y_tr.quake_number < 16)].index\nX_tr, y_tr = X_tr.iloc[keep_idx], y_tr.iloc[keep_idx]\ny_tr.quake_number = y_tr.quake_number - 1  # start counting at 0\nplt.figure(figsize=(10, 5))\nplt.title(\"New count (y_tr)\")\nax = sns.countplot(x=\"quake_number\", data=y_tr, palette='GnBu_d')","execution_count":6,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 720x360 with 1 Axes>","image/png":"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\n"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"<h2>Predict earthquake (multiclass model)</h2>\n\nUsing a classifier with 15 possible classes:"},{"metadata":{"trusted":true},"cell_type":"code","source":"params = {\n    'objective': 'multiclass',  # Softmax\n    'metric': ['multi_logloss', 'multi_error'],\n    'num_class': 15,\n    \"boosting\": \"gbdt\",\n    'num_leaves': 32,\n    'min_data_in_leaf': 10, \n    'max_depth': -1,\n    'learning_rate': 0.01,\n    \"feature_fraction\": 1,\n    \"bagging_freq\": 5,\n    \"bagging_fraction\": 0.9,\n    \"bagging_seed\": 19,\n    \"lambda_l2\": 0.1,\n    \"num_boost_round\": 90000,\n    \"verbosity\": -1,\n    \"nthread\": -1,\n}\ndataset = lgb.Dataset(X_tr, label=y_tr.quake_number)\nresult = lgb.cv(params, dataset, nfold=10, early_stopping_rounds=100, stratified=False)\nplot_multiclass(result)","execution_count":7,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 720x360 with 2 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAqIAAAFmCAYAAAC2kPjMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xd4FVX6wPHvbbktCTUBaSLtSFGx664ooquigooN3cWCBRuuBdayq6Kiq6CCDcWOvfeC6Kqr/uwrYAEORTpICaTdXub3x0zgElNuIMkk5P08T54kU868czI3971nzjnjMAwDIYQQQgghGpvT7gCEEEIIIUTLJImoEEIIIYSwhSSiQgghhBDCFpKICiGEEEIIW0giKoQQQgghbCGJqBBCCCGEsIXb7gCEaMqUUk8B7bXWx9djmYOBT4ECrfXG+iq3jjF4gF+AC7TWnyulugNLgf211j9Us0+t22xnLJMBn9Z6bH2VWcfj7wd8D+ymtV5mRwzNkVLqJmBXrfVou2OpL0qpS4FjtNbD7I5FiJZCWkSFaJkuBNZorT+3OxBgEnC2UqqH3YGI7CilCoGrgYl2x1LPHgP2VUoNsjsQIVoKSUSFaGGUUg7gcuBxu2MB0FpvAGYBF9sdS3OilMqx8fDnA99prX/bkUKslvkmQ2sdA57HfH0IIRqB3JoXog6UUl7gTuAMoBUwBxintf4yY5vjgHuAXYHvgGnAC9Rw61cpNQK4GegDrAceBm7XWhsZ6ycAvYEI8DNwmtZ6nVKqK/AAMAjwASuACVrrF6s5jX2tct6tYl0fpdRUYD9gGXC51npWNTEPplIXg6pu3yul+gGTgUOt2P8DXKm1/j2juLeB24Hx1cSMUuoY4J/AAMDAvJ1+hdZ6fqVjnwJcBPzZOoe/a60/qlTOVKC7VcZD1R0zY5+/AX8HdrfO4b/WsVdnbLM7ZuvuYYAL8290odb6Z2v92cA4zL9xMTBTa322tc4ATtVav5pR3jLgAa31XRnbXAYcARwNPKSUugZ4BBgCdARWAY8Cd2mt0xllVXlspdQTQGFm1xOllNOqt6la63uqqZIzreNk1lGNr42M6+U4zGt5IDBCKTUf8/VyIJAHaOBGrfW7GWUvw2yt7GqVXwrcq7WenLFNHyumA4HlwJXAy8BlWuunrG06A3db9QfwFebfcVHGqbwNfKSUCmitw9WcvxCinkiLqBB1Mwk4HRgN7I2ZbMxUSu0CoJTqBrwOvAfsBdxn7VMtpdS+wCvWfnsA1wLXYSYdKKU6Ai8CM4C+mAndMxlFTAMCwOFAf+AKzGSjOoOAJVrrqraZZMU8EPgIeMt6894uVr18jtkf9QDgSCDXKjfz/893QGelVM8aigtiJpAHAIOBEuCdKloGb7POYS/MRPNFpVSuFU9X4E3r3AYC91PL38eSA9xklXk80B7zw0XFeXYCvsRMkP8C7AM8iJmQopQaA0wHngT2BI616qSubgLex7xOHsT8H74aOA3z2vgncD1wbkZsNR37UeCYiuvX8hfMpDbzGttCKdUW6AdU7idc42sjw53AvzCT+m8xr4cPrOPuBbwGvG4l9pmutMrcxypjklLqYCsmJ/AGkAQOAs6x6sqbEXcAMxGOYn5YOBhYC3xsravwA2YjzcFVnb8Qon5Ji6gQWVJKBTFvH5+vtX7PWnYRZmvUpZhvrhcDv2mtr7J201ZLzW01FH0V8F+t9U3W7wuVUr2BazATpU6AB3hVa73c2iYzidkVeE1rPdf6fWktp7IrsKaadQ9prV+2zu3vmC1HF1vntj0uBuZqra+pWKCUOgvYhNnq+p21uCKe7sCSqgrSWr+W+btS6lzMlrEDMJPAClO01u9Y21wPnIWZdH5pxbMCs6XXABZYf59bazoJrfUTGb/+ppS6GJivlOqitV6F+fcPYbZqxq3tFmbscwN/bGH8X03HrMZLWuvHKi27MePnZUqpfTBbDSu6XlR7bK3110qpBcDZwB3WutHA21aXiap0AxxkXENZvjYqTKjUyr4BmJvx+21KqWGYLduZfVBnaa0fsH6+Xyl1OWbr8NeYSawCjqpopVZKXQn8X8b+I624z8240zAG8w7E8Zitp2itw0qpEsxrUQjRwCQRFSJ7PTETwi1vblrrlFLqa8wWIjBbeb6vtN+3tZTbF7MFNdOXwE1KqXzMN+mPgV+UUrOsn1/NSBTuBR62bjn/B3hDa11TkuPHbBWqytcVP2it00qpb9l6bttjX+BQpVR5Fet6sjURjWTEViWrtfRWzFuvBZitgU7MxCjTTxk/VyRLhdb3vsA3FYmI5WtqYSV3N2EmtG0xExqsY6/CbAH8MiMJzdy3EOiM+bfZUX+YrcBK+M7H/IDhx7xGl9fh2I8ClwB3WK2dJwAn1bB9xd8o8xrK5rVR5TlYSexNmMngLlY5Prb9O1LF72vY+nfdHXPw3eqM9d8D6Yzf9wV2A8qUUpnlBKz4M0Wo4VoUQtQfuTUvRP0wat9k+8rVWqeAo6yvn4DzgEVKqb0AtNaPY77BPonZB/ArpdSEGsrcCLSph9gq3uQdGcsqDz5xYibZAyt9Ve6j2tb6Xl0rHNb2BcAYzGR0b8xbsZVvzScqfshIOLf7f52VKH0IhIFRwP7AMdbq+howZLBtPcIf6xLMVtfM2E7H7K7wFGbr9UDMrhp1iesZYFel1CHAXzH/Bh/WsH3FlGPZXkOVXxuhSr/fBZyK2XJ7GOY5fEcNf9eMcuvyd3Vi9lutfC32wey6kKktNV+LQoh6IomoENlbAsQxB8EAoJRyYfYlm2ctWoB5yznTAbWUOz+zTMshwCqtdRmYCZXW+mut9c2YidAazP54WOtXaa0f0Vqfhnmr9sIajjfbDF1V9fo/KOPcHFbs86spp+KNOrMP4MBK2/yI2W91udZ6caWvsoztBmAmGj9XdSClVDvMVq/btdYfWwOU8qj7XZ35wIHWuVU4qLqNLbtj9gm9Xmv9udZ6AVtb4irMBg6paiS71no9Zj/OI2o4xgYy6lEp1YFt67U6hwDfaq0f0Fr/qLVeTEbrXjbH1lpvwuyfPNr6mpE50KkKSzC7RPSrtKy210ZN5/C01vo1rfVPmC3MNfUVrsoCoJPVV7fCfmz7Hvcj0AvYWMW1uCkj7p6YLbI/1jEGIcR2kFvzQmRJax1SSj0E3KmU2ojZF/NKoANmKxSYo92vUkrdhXnLsz9mCx5U32p6N/C91Yr5PGaieTXmoBOUUgdhDvL5EFiH2RLYFesNXil1L+Zgj4VAPmZrXU1v/p9ivtHuidlClOlipdRCzITwEszbvdWNKl8MrAQmKKWuxexTV7kv6YPABcBLSqk7MROuHpiDa67OSEYHAV/UMEp5M2ZL3AVKqZWYt5snY7aI1sXDmHU7VSk1DXPQz0W17LMCiAGXKaUexLy9X7lP6TSrnJeVUrdZ8e4PzNdaz8HsIzxFKbUOs4U4AByhtb7b2v8T4FKl1FdACnMGgeq6T2RaCJyjlBqK+fcYidmquDljm9qODea1OhOzFfbkmg5oddn4GDOBfNVals1ro6ZzOEkp9Rbmh5GbMK/PuvgIc7T9DKXUOMzb6vdgXh8Vr7vnMGcOeEspdSPm37UrZleEhzNGzg/C7OedOZJeCNFApEVUiLq5BngJ8zb4HMxk7hit9VoAazDRycBwzL6dV2JOywTVJBZa6x8xb02ejDkI6Q7rq2JgRglmS9O7wCLMxPVWrfWz1non5qCmeZhvyOswB59USWtdhNkC9tcqVl+LOXhqLmZCe5I1GKeqchKYiU8Pa/ubsZLnjG3WWLGnMROdXzGT05j1VeEMKk0HVKmcNGYL8J6YdfQg5q3cWHX7VFPOCmCEdW4Vf59ra9lnA2Z9nohZxzdh1lHmNqsxZzPIwUz0ZwNjsRJlrfVDmIN2LrDin4n5IaXC1cBvwGeYyd1jmINoajMdc5DN85h9IrtjXh+ZsdV2bKzjrgI+y3Ju0EeA061Wzwo1vjZqcBXmuX6B+YHqG+vnrFnXx0mYo+S/w5xh4jbMJDRqbRPG/Bv9hjlLxQJruzZsm7jXeC0KIeqXwzAaqmubEAK2jD6/BWhdaZCMbZRS/TETpl5a61KbYzkOs3VzT611XVs4RT1QSvkxb+GP1Vo/l+U+XwPTtNZVTvNkN6sP9Rxgv1oG72XuMwBzYFcfrXVJQ8YnhDDJrXkh6pkyn1f9PeZt6IMwW+6eaipJKIDW+lfrFuZubDt1jh2CmFPqSBLayKx+wu0xJ+uPYE1hlKUxmN1EmgSl1EmYA6EWYbYM34N5bdelr2cn4CxJQoVoPNIiKkQ9U0pNwewD2Q7zdueLwC1VTe0jhJ3U1qdRrQLOq+4pWs2BNT/tvzD7fW7G7G5wpdZ6nZ1xCSFqJomoEEIIIYSwhQxWEkIIIYQQtpBEVAghhBBC2EISUSGEEEIIYQtJRIUQQgghhC0kERVCCCGEELaQRFQIIYQQQthCElEhhBBCCGELSUSFEEIIIYQtJBEVQgghhBC2kERUCCGEEELYQhJRIYQQQghhC0lEhRBCCCGELSQRFUIIIYQQtpBEVAghhBBC2EISUSGEEEIIYQtJRIUQQgghhC0kERVCCCGEELaQRFQIIYQQQthCElEhhBBCCGGLlpKIuoHu1nchhBBCiKasxeQtO/0JWnoD84DTgTU2xyKEEEIIUZNOwEtAP2C+zbE0qJaSiA62vr9kZxBCCCGEEHUwGElEdwoLAEpLI6RS6QY7SJs2QTZvDjVY+TsTqau6kfrKntRV9qSusid1lT2pq+xVV1cul5P8fD9Y+cvOrKUkogmAVCpNMtlwiSjQ4OXvTKSu6kbqK3tSV9mTusqe1FX2pK6yV0tdJRorDru0lMFKQgghhBCiiZFEVAghhBBC2EISUSGEEEIIYQtJRIUQQgghhC0kERVCCCGEELaQRFQIIYQQQthCElEhhBBCCGGLljKPqBBCCCHETkkp1QeYAbQDioCztNaLKm1zAzASSGHOT3q91vpDa10AeBLYF0gC47TW7zZG7NIiWk9+/NHJEUfAxo0Ou0MRQgghRAPwbPqc1t8diTO8xO5QKnsYeFBr3Qd4EJhexTbfAftrrfcERgMvKaX81rpxQKnWuhcwDHhMKZXbCHFLIlpf8vLgyy9h3DgvhmF3NEKIPzAMSO/0DykRQjQEw8C/YhqtfjwBR7IEw9PO7oi2UEoVAvsAL1iLXgD2UUoVZG6ntf5Qax22fv0JcGC2oAKcjpW8Wi2pPwBDGzh0QBLRetO7d5qJE+H99z289JL0eBCiSTDSuIu/Iaivp+2Xe9L+007kLrgaZ2Sl3ZEJIeyWLMdd8iPetS/iW/UkjkRJ1dulIuT9ehG5+lri7YdSfMAnGJ7WjRLiBx980FEp1b3SV+WDdwVWa61TANb3Ndby6pwFLNFar7J+7wYsz1i/opb9602LypjatAk2aPnjx5tf4K9tUwEUFOTZHUKzIvVVSaIcShdA6XwomQ+l86BUQyrClmaARBnEN4EzBzr+BbyD8S9/Cv+qJ2G3UdDhcHOfknlQvhg6HAF73gqeRrkj1STIdZW9ZlVX4TVQ8uu2y9rsDb72jXL4KusqnYLNP0K8eOsyhwvyekKgKzistrHQSlj1Bqx6E1xB6DceCg+tv+A2fgdfj4Kyhdsszlt8I/QZC+rv4G1n/l9Y+Tose9bcdo9b8A74J15H/bbh1XRdTZ48+ZUqFt8MTNje4ymlDgNuBf6yvWXUpxaViG7eHCKZTDdY+QUFefz4YzmHHRZkwIAUb7wRweVqsMM1awUFeWzYUGZ3GM2G1BdgGOSsfxv/6hm4QhpXdGurpuFwkwr0IBXogze3HdGoeQvecHpItDmUePujMDytAHB2uQb/snvxL5uB47cnMRwuUv4epH2d8Oh7SS9/jbK+U0m0bxL/oxuUXFfZq7GuUhEc6dh2lWs43OCuhw8+hoEzshTv+nfxrn8LT8n3f9gknVNAyZ7Pkmxz8I4frwbb1FU6gWfzF3jXvY13w7s44+ur3MdwBkgG+4DDgad0NgDJYF+ciY04Pz6MeOs/Ed5tHIl2R4Bj+8diuDd/TavZp2B42hLt+S+SQUUqqHCkygksm4r314kYC6aQ8nbEbfUDTbQ6kPDAl4kXHAMbQ9t97KpUd1253U7atAkyfvz4U6+44oofKq0urvT7SqCzUsqltU4ppVxAJ2v5NpRSBwPPAidorXXGqhXArsAG6/duwKfbdVJ15DBaRofGQ4AvGiMR3bChjJdecjN2rJ9//SvG5ZfHG+x4zZm8AdZNS68vZ3QtuQuuxrvhXVL+3Ui02o9UcPctbyKpQA9weoDs68oRL8IZ30gqsJvZYgq4i78lb95luEOaaIeTSeYPrHZ/w51PKtiHZFBh5DROK1N9a+nXVV1UWVfpGIHfJhNYNgWHsf39j6O7nE55nzswcrLvd+hIluJd+yLu0p9whzSukMaZNPOTRN5A4oXDSLQ+GMPptrYvI1f/A1dkOeVqEtEu5+1QQleTgjZuSvTbeNe/Rc6GD3AmizFcQWLtjyJecDwp/9Y7vo50DFd4Ca6Qxh3SOFIRYu2PJl44jFSwN6TC+FY/TWDZvbhiqzFcQZKBPqRy+5DM7U+081kYnrZZxeXZ9F9azT6dlK8zJfu+Q9rX6Q/buMrnE1g2FWd8A7GCY4kXHk/a27He6qay2hJRYBDwZW3lKKU+Ax7TWj+rlPobcJ7W+vBK2+wPvAacqrX+ttK6CUBnrfUFSqne1jF7aa0b/B+EJKL1qOKCMgw4/3wfH3zg5t13w+yzT8Mds7mSN8C6abH1lY7hW/McwUU34kjHCfX8J5Ful4Kz+ps5O1xX6RiBpXcRWDoFh5HdB8m0px3h3cYR2fXS7T+uDaqtKyONK7SAVLBvgyUrzU3lunIXf2N9aFlItOMpJPP33a5yndFV+Fc+guHOp1zdSazjqWadp+O4wr/hSEdIBnpvaTV1xIvwr3gI/8pHcCaLSXvab/lAlszrR7z90aT9u1Z5LEeimLxfzse7cRaRTmcR2+X0revSEVzhxbjKF5pJbWztNvumAz2IFQ4nVngcRk6B1dL5f3jXv41n8xfbDAR0J9ZBMkTa3Zp4wVBihcOJtxsCrh3otpaO4133Ou6SH63EeyGu2GrSOYWU7T6ZeOGJW65Vd+lsAkun4Iyu2Pph0RUkd9ENpAI9Kd7nbQxv4fbHUo/qMRHdHXP6pjbAZszpm7RS6n3gRq31D0qp74HuwOqMXUdprX9WSgWBp4C9Mad3+ofW+q0dObdsSSJajzIvqOJiOOKIIA4HfPJJiPz8Bjtss9RiE6vt1JzqyxVaiG/VU0Q7nUkqb0DdC0iWk1P0Md71b5Oz4UOcqTLibQ6lrN+9pAM9a9293uoqHYcaWrmc8aItrTg5Gz8iZ9OnlPZ/iFinv+74sRtJlXVlGOTq8fhXPkIydwDh3a4m1uFEsy9fc2Okt/Y73EFb6iodI7jwBgIrHybl61ov3Thc5fPIm3cZnpIfSOTtjSNVjiuyFIeR3LJNyteFVKAXnpLvcaRCxAqHE97tapL5e9ftYEaKwJLbCC69q8rVaXdrUrm7k/J23lp3RhpP6Y+4IksxcJJstQ+u8BKcic0YzgDxtodiuLf2c/Tld6A490gSbQZtuVPREFxlP5H362V4yuYQKziOaOez8a98hJyij0m7W5PM3wtXaBGu2BoAEnl7UbLPm3VqeW5o9ZWINmeSiNajyhfU9987GT48wHHHJXn00ag0LGRoTolVU9Ac6stV9hOBpXfjXfcmDgySgZ5sPuhLc7BBdVIhcjb+B0/pD7jKzduLrsgyHKRJe9oRKzyeWOFwEu2OzLplzpa6SsdpNftUPJu/oGTgyyTaH9m4x99OVdWVf9lUchfdSKzwRFyhebhDC0kGehHt9Fdw+rZsl/a0NVvhgn3qp48j4CrXeIq/Iu3tRDJXkfZ1275EMhUhb8FV5GycRcnAF0m22n+HYysoyKNo1SLy547CU/Itka4XUt5rQr2dO0YK/8rpeNe8SNrf1Wrl7IPhDOAOL8RVvgBXeBGpYB/C3a8ildt3hw7nKp+PM75x6+GdHlL+HmZrZ1WvNcPAVf6r+QGx6GNSgV5WS+cRf2jpbNTXYDqJf8U0gktuw5GOkPa0J7zrZUS7no/hNluAHMlSnJFlpAJ9wOWrpcDGJYmoJKL1qqoL6v77c7j1Vi+TJ0c5+2yZw7BCc0ismpImXV/pOHnzLsW39iXS7nwiXS8kmbcX+T+dRbTLeZT3vWfb7VMRvOvfwrv+HXI2fowjHcFweEgFem1JbBJtDyPR+uAab8FXx666ciRLafXDsbjDSyje74Ma+5c2FZXryrv2JfJ/uYBohxGU7fEEYJCz/h0CS+/CU/ZTteWkfF0wXFsTMsPpJdF2MLHCYSRb7Vd9Mpksx10+j5yNM/Gufxt3aNtRzIbTTzKvP7GC44gXDjf7DNbCGV1F/ty/4imdTTqnAFIRSvd+hUSbP9e679YTCuEKLyHt774lmSlw/ErqvyNwJkoo6z+NWMcR2ZfXwtjxGnSGl+IpnU2s4BhwBRr12DtCElFJROtVVRdUOg1nnOHnq69czJwZpn9/6S8KTTyxaoKabH2lIuT/NArvxlmEdhtHZNfLt8yvF9TXE1jxAMV7v76lhdAZXkqruWfiLv+VlHcX4oXDzBbP1gfX2y08O+vKGV1L6+//giMdId5u6+1aw+klFei5pb8aTp/Z+hvSuMOLMFyBLS1gqaDakvw0CMPAXfo/cja8T9BrUObcjWRQ4YwXkf/TKBKtD6Jkn9fB6d1mH0eyFKh4vzBwxtZv6ZrgCi/CkYpu2dyR3Ixn81c4jAQpbyczCXR4tu4bX2f28YuuqlhCou0gYoXDSLQ9HEd845YBOJ7ib/CU/g8wR1GbCf7WFruUr9OWenMkSsj/5TxIRSkb8CjJ/IG0+nE4rshKSga+SKLdNmM3tuFIbCZnw0zzA1LRxzjS5vmkvJ1IBXuTU/w1KW8nSvZ6fvu6nLQgTfb/VRMkiagkovWqugtqwwYHQ4YECARg1qwQrVo1WAjNhvyjqpsmWV+pEK3mjMSz6XPK+95LtMs5ldZHafPtoTgSxWz+0ze4S+eQ/5O5TdmA6cTbH11v/fcy2V1XrtBC8n6+AGeiaMsyRyq0ze+Z0q48HOnYNgOjzOTHbB1O5fYn3v6oKkf4ZsUa9OIKaWtwyTvm6GOHG4fDBRnTDiVz+1G838x6mazbkSg2WzrXvY277Ge2JrEVt/X7WOeoSLT5c4399pzRVeSsf9eMPZI553YaZ2wtDiO19RwCvSjd6wVSucqMI76B1v8bjiu8mLK+U4l1GLHNrWRneCmBZVPxrXkOhxEn5e1stuS2PgBnZAXu0AKz7lr1ZGOPyU2qf2FTZfdrsDmRRFQS0XpV04vvm29cjBjhZ8iQFE8/HcHZwp9pJf+o6qZJ1ZdhmLc/fzkfd/G3lPV/iFinM6rc1F06l9bfHU4quDuu8nmkcvtSstdzpAM9Giy8JlVXGRyJTbhCC3GXa0hHzWmngsqcGsZI4YosNVsJK1oZQwtwhxbiSJnzFiZaHWCNWh5GOrBbzcdKluFb9Ti+Nc/hCi/ZMujFcPqItzuCWOEw4gXH0L5jF4pW/ow7tBBndDWxwuFNZjRx1jISbWd8A7GOp/whkXbEi2g1ewSe0tnm4Jr2RxErOJqcTf/F+/srgIto51FEO/3VHP1eRR/JpnpdNUVSV9mTRFQS0XpV24vv8cc9XHedj/HjY4wf37LnF5V/VHXTUPWVmRy5wotI5xRYic7WRNGRLCVn4yxyiv6Dq3w+rtAinKkyDIebsgGP1dpXLvDbZIJLbiVWeCKl/afV3+COauxU15Zh4AotMG8Xr38HT9lcABJ5e1rdGk4gFVRbEidHYhP+FdPxr3gIZ7KYeJtBJFofuE3rY2b/uZ2qrmqTTuDZ/CXe9W/jXf8Ozvh6DGeASJfRRHYdS9q3S427t6i62kFSV9mTRFQS0XpV24vPMGDsWB8vv+zh2WfDHHVUqtptd3byj6pu6rW+DIOcoo/MASjF32xd7PRueTpMMncA8fZH4Sr/lZyiT3EYcXO+wrw9zNvFQUWizZ9I5fbL6niu8p9J5e7RKHNS7szXljOyDO+6d8y5G0u+rXa7WMFxhHcbR7JVzXNb7sx1VSMjhbvsJ1K+blnfam+xdbUdpK6yJ4loC3vEp90cDpg8Ocr8+U4uucTPrFkhevRoER8EhA0c8SJyF1yFwzCsvoa7g2HgX34/nrI5pHxdCPW8wUouFWl/N5zRVdYjAt/Gv2wKaV9XIl0vJNZhOMlWB2xfn06Hg1TenvV/gi1Q2t+dSPexRLqPxRldS87GD3DGft+6gcNFrOB4Unn97QuyOXC46j7/phCiQUgi2sj8fnjyyQhHHRXg3HP9vPdemNyGvVMpdmI569/DGV9HtPO527Y2GinyfzkPz6YvSfm6kLP+bRyYdwOS/h6U9XuQ6C6nb3m0ZYW0f1ciu15qPiEoFQJnQJ6s00SlfbsQ7TLa7jCEEGKHSCJqg27dDB5+OMrIkX6uvNLHI4/IZPeibpzRteTqcXjXvwOYI7VDff69JWkMLPk3OUWfUNb3XqJdzoVUdMuTUBKtD8pufs6aJqIXQggh6oEkojYZPDjF9dfHmTjRy957p7jkEpnsXlTNGVkJxQaucnP0tKf4a4KLbsKRjlHe62acsbUEVkwzf9/9bnI2ziK4dBKRTn8j2vkcsxCXj1Ref1pur2QhhBBNkSSiNho7Ns6cOU5uucXLHnukGTRI0gSxlXvz1wSXTiKn6D8AtM1YF28ziPK+95IK9jJHwbn8BJZNwZHYRE7RpyTy9qJ897vltroQQogmTRJRGzkccN99URYtCnDhhT7efz/MbrvJ4KWdmpEm8NskfKueIO3rvGVKHXPqGGsKnnQc79oXyNn8JWlPe0I9/0Vwlz0pKY2YRbhbkWg7eOvAIYeDUK8JGE4fwd/+TdrdmtI9n/7D85+FEEKIpkYSUZvl5sKMGRGOPTbAyJEB3nsvTPv2kozujBzJUvJ+GYN3w3vE2x4OpPEUfYJv7fN/2Dbl3YVydQeRzufbWtYgAAAgAElEQVSAK0CwII94TdOhOByEe15nzhfp717rhOdCCCFEUyCJaBPQo4fBM89EOPnkAKNG+XnttTCBQO37iebDFVpE/twzcYUXmwlm14szJiEvxhnfuM32KX+3P4xoz0as48n1Eq8QQgjRGFr4gyabjv33T/Pww1Fmz3YyZoyfZNLuiER9ydkwk9bfDcEZ30DJPm8R6XbJNn03DU9rUsFe23xtTxIqhBBCNDeSiDYhxx6b5PbbY3z4oZtrrvHSMh56tRMzDAK/TSZ/zumk/Luy+cD/kmh7qN1RCSGEEE2G3JpvYkaPTrB2rYN77/XSubPBVVe17GfSN1vJcvJ/vRjv+reIdjyFsn4PbPOMbyGEEEJIItokXX99nDVrnNxxh5dOndKMHCn36ZsTR6KYVv87HnfZL5T3vo3IrpfJNEpCCCFEFRolEVVKtQOeAXoCcWARMEZrvaGKbccClwIJIKW1Hmgtfwo4EqgY1fGK1vq2ho++8TkcMGVKlHXrHFx1lY/CwghDhsgco81COkb+3L/iLp9P6cCXiBccbXdEQgghRJPVWH1EDWCS1lpprfcAlgB3VN5IKTUCOBXY39qu8rv4HVrrgdbXTpmEVsjJMZ9Jv/vuaUaP9jN3rnTnbfKMNHm/XETO5i8o6z9NklAhhBCiFo3SIqq13gR8lrHoG+DiKja9GrhBa11m7beursdSSrUGWmcumz59esfBgwfXtSjb5eXB889HOO64AKef7uf11yP065e2OyxRjeCiG/Cte43y3rcQ2+V0u8MRQgghmjyH0chDs5VSTmAW8LbW+r5K6zYDk4HjgRxgutb6UWvdU8ChQAizRfU6rfX8KsqfANyUuWyfffbhhRdeqPdzaSxLlsBhh0E8Dp99Bv362R1RC1CyABbeD0YaWvWF/L7Qqh/4O23b39NIw8ZvYekMWDwd+lwG+94nfUKFEELUh0HAl3YH0ZDsGKx0P1AOPFDFOhfQFTgEaA/8n1JKa60/B/4JrNVap5VSZwEzlVI9tNaVO09OBZ7KXDBmzJj9gFc2bw6RTDZci2JBQR4banr6zXbKz4dXX3Vw4okBBg+Gt94K06tX857bqaHqake5yn4isPRuvOveBKcPw+nFmSzesj7tyjOfXhRUGE4vORs+wBX/HcPhIdr5XMq73Qoby+s9rqZaX02R1FX2pK6yJ3WVPamr7FVXV263kzZtgjZE1PgaNRFVSt0F9AaGaa2ryghXAC9Y69YrpT4CDgA+11qvrthIa/20UmoK0AVYnlmA1roYKGZbXerxNGzRs6fB669HOOEEPyNGBHjzzTA9ejTvZLSpCSy+leDSyaTd+YR3u5pIt0swPO1wxtfhCmlcIY07pHGFFuIp+gRnspR4+yMJFQ4n3v5oDE8ru09BCCGEaFYaLRFVSt0O7Ascp7WOVbPZ88AxwOdKqSBmk/Qb1v6dK5JRpdTRQApYXU05O6XevdO89lqEESO2JqPdu0syWh8csfUElt1LrPAEyvrdj+HZ2s047e1I2tuRRNvDtt3JMOQWvBBCCLEDGmUotlKqP3Ad0An4Sik1RylVkWDOUUp1sjadAnRVSv0KfAc8q7X+yFo3Qyn1s1JqLvAvYLjWusVNsNm3b5pXXokQiTg4+eQAK1dKIlQf/KseBSNBqNeN2yShNZIkVAghhNghjTVq/legynftinlCrZ8jwKhqtjuyYaJrfgYMSPPKK2FOPjnAiBEB3norTKdO0jK63VIR/CsfI14wlFSwt93RCCGEEC2GTE7ZTO25Z5qXXw6zaZODk04KsHattM5tL9/al3Amioh0u9TuUIQQQogWRRLRZmzvvdO8+GKYDRscnHBCgFWrJBmtM8PAv+JBEnl7kWhziN3RCCGEEC2KJKLN3P77m7fpi4rM6Z1WrJBktC48RR/jDmkiu14qfT6FEEKIRiaJ6E5g333TvPZamJISMxldulQSqmwFlj9AKqcjsQ4j7A5FCCGEaHEkEd1JDByY5vXXw4RCDkaMkNv02XCV/UrOpk+JdBsDzhy7wxFCCCFaHElEdyJ77JHm1VfDlJU5OPXUAOvXSzJaLSNNrr6GtCuXaJdz7Y5GCCGEaJEkEd3J7LFHmueei7BmjYPTT/dTUmJ3RE2Tf8U0cjZ/TkjdgeFpa3c4QgghRIskiehO6MADUzz5ZISFC52ceWaAUMjuiGxiGHg2foQzumqbxa7y+QQX30ys/VCinaqctlYIIYQQjUAS0Z3UkCEpHn44yv/+5+Tcc/3Eqnuo6k7KGVlGq9kn0Xr2ybT9an/8Kx4CIwXpOHm/XIjhyqOs3/0yUl4IIYSwkSSiO7Fhw5Lcc0+Uzz5zc8klPlIpuyNqBEYK//IHaPvVQbiLv6O8920kWh9Mrr6G1t8fRe6Cq/GUzaWs370Y3kK7oxVCCCFatEZ5xKewz5lnJiktjXLjjT6uvtpgypTYztEIaBg4Y2txhRbgDmlcoYW4Qhp3+XyciSJi7Y+mvO8U0r4uRHa9DO/vL5Orr8FT8j3RXc4kXjjM7jMQQgghWjxJRFuAiy5KUFzs4J57vOTlwS23NN9k1BleSt68S3GXzsWZKtuyPO1uTSqoiBUcR7z9UWaiWXGSDgexXU4n3u4IvL+/SqzTX22KXgghhBCZJBFtIa65Jk5pqYPp03PIzzcYPz5ud0h1Z6TI/+VCXKH5xDqdQTKoSAUVyaDCyCmstb+nkdOeaLeLGilYIYQQQtRGEtEWwuGAiRNjlJc7mDzZSyBgcOmlCbvDqhP/sql4Sr6ldMCjxHY53e5whBBCiCZBKdUHmAG0A4qAs7TWiyptcxRwO7AHcL/WelzGukLgSaAr4AE+BS7XWicbOnYZrNSCOJ0wZUqUE05IcPPNPp54wmN3SFlzlf1EcMntRDucRKzjaXaHI4QQQjQlDwMPaq37AA8C06vY5jfgfGByFeuuB+ZrrfcE9gT2BRrl2dfSItrCuFwwbVqUaNTBtdf6CAQMRo5s8A88OyYVJf/nC0h72lG++z0y5ZIQQogW4YMPPuh4xRVXdK+0uFhrXVzxi9WauQ/wF2vRC8ADSqkCrfWGiu201out7U+s4lAGkKeUcgJeIAdYXW8nUoMWlYi2aRNs8GMUFOQ1+DHqw8yZFT/5bYsh67r68WYIzYfB79O+U/cGjakpay7XVlMgdZU9qavsSV1lT+oqezXV1eTJk1+pYvHNwISM37sCq7XWKQCtdUoptcZavuGPu1fpVuA1YC0QBB7QWv9flvvukBaViG7eHCKZTDdY+QUFeWzYUFb7hk1EOAyjRvn58ksX994bbdSW0Wzryr/8QXIX3k2ky3mUew6BZlS/9am5XVt2krrKntRV9qSusid1lb3q6srtdtKmTZDx48efesUVV/xQaXXxH3bYcacCPwFHAHnAB0qpU7TWrzbAsbbRohJRsa1AAJ55JsJZZ/n5+999pNNRzjyz6dym9y+9m9zFNxMrPIFydafd4QghhBCNaujQob8PHTp0WS2brQQ6K6VcVmuoC+hkLc/WWGC01joNlCil3gIOBxo8EZXBSi1cRTJ62GEprrjCz7PPNoEBTIZBYPFEchffTLTjaZTu8SQ4c+yOSgghhGhytNbrgTnAGdaiM4DZmf1Ds7AUOAZAKZUDHAn8Up9xVkcSUYHfD08/HeGII5JcdZWPGTPsTUYDv91JcOkkIp3PpmzAdHBKw70QQghRg4uAsUqphZitmxcBKKXeV0rtZ/18iFJqFXAVMEYptUopdbS1/xXAIKXUz5hJ7ULg0cYIXN7hBQA+Hzz1VITRo/2MH28+l3706MafZ9QR30Bg2d1EO4ygvO+94JDPSkIIIURNtNYLgAOrWH5sxs9fAl2q2X8JW0fdNyp5lxdbeL3wxBMRjj46ybXX+njsscZvGfWvfBxHOka45/WShAohhBA7OXmnF9vweuHxxyMMHZrg+ut9TJvWiMloKop/1aPE2h9NKtin8Y4rhBBCCFtIIir+ICcHHnssyvDhCSZM8HH77TkYRsMf1/f7KzjjG4h0u7ThDyaEEEII20kfUVEljwemT4/SqpXB1KleioocTJoUw+VqoAMaBv4VD5LMHUCi7WENdBAhhBBCNCWSiIpquVxw110x2rUzk9HNmx089FAUr7f+j+XZ9Cnu8nmU9n9IHuEphBBCtBCSiIoaORxw/fVx2rQxuOkmH5s2OXjqqQitW9fvcQLLHyCV04FYx1Pqt2AhhBBCNFnSR1Rk5eKLEzz0UITvv3dx/PEBVqyov1ZLV/kCcoo+Jtr1AnA2QHOrEEIIIZokSURF1k4+OcnLL0dYt87J0KEB5syph8vHMAguugnD6SfS5bwdL08IIYQQzYYkoqJO/vznFO++G8bngxNPDDBr1o6NXvKteQbvxg8I9boBI6ddPUUphBBCiOZAElFRZ0qlef/9ML17pznrLD9PPrl9c406w0sJ6muJtzmUSLdL6jlKIYQQQjR1koiK7dKhg8Ebb4Q58sgU11zj45Zbckin61BAOkX+rxeBw0lZ/4fkKUpCCCFECyTv/mK75eaaz6c/55w4DzzgZcwYH5FIljsvuBtP8deUq0mk/V0bNE4hhBBCNE0yfZPYIW433HlnjK5dDSZOzGHlygAzZkTo0KH6RzG5S/4HP91ArHA4sV3OaMRohRBCCNGUSIuo2GEOB4wdG+fJJ6MsWODk6KMD/Pxz1ZeWM7yUVnNOA/8ulPWdKpPXCyGEEC2YJKKi3hx7bJJ33gkDMGxYgPff37bB3RHfSKvZI8BIwuCZGDnt7QhTCCGEEE2EJKKiXu2xR5oPPwyz++5pzj3Xx3335WAYQCpMqzmn4YqupmTgy9Bqd7tDFUIIIYTNJBEV9a5iRP2JJyaZONHLFZc7yZ1zLu6SHynd4wmSrQ+0O0QhhBBCNAEyWEk0CL8fHn44St8+YQ5MjMK/6QPWdpmCu/B4u0MTQgghRBMhLaKiwTjSYW46fATD932Hy595kD+dfTm//SaDk4QQQghhkkRUNAhHsoxWs0/BU/QZpf0fYujlZ1NWBscdF+CHH+SyE0IIIYQkoqKB5P10Dp7irynb4zFinf7Kfvulee+9MHl5MGJEgDfftDtCIYQQQthNElFR79zF3+It+ohQrwnEOp6yZXmPHgbvvx+mf/80I0bA449v3zPqhRBCCLFzaJTBSkqpdsAzQE8gDiwCxmitN1Sx7VjgUiABpLTWA63lAeBJYF8gCYzTWr/bGPGLugksvYu0px2Rruf/YV379gavvRbm73/P47rrfKxc6eTGG2M45SOREEII0eI01tu/AUzSWiut9R7AEuCOyhsppUYApwL7W9sdnbF6HFCqte4FDAMeU0rlNnzooi5cZT/h3fghkW6XgCtY5TaBALz2GoweHWfatBwuushHNNrIgQohhBDCdo3SIqq13gR8lrHoG+DiKja9GrhBa11m7bcuY93pwNnW8kVKqR+AocArmQUopVoDrTOXTZ8+vePgwYN37CREVgJL7yHtzifS9YIat3O54N//jtGlS5pbbvGxdq2DJ5+M0r599c+oF0IIIcTOpdHnEVVKOTGT0LerWN0POEgpNRHIAaZrrR+11nUDlmdsuwLoWkUZVwA3ZS6YPn06gwcPpk2bqlvo6lNBQV6DH6PJKtWw7g3ody3tO1X1p9lWYWEeN98M/frBOee4GTo0l7fegr32aoRYm6EWfW3VkdRV9qSusid1lT2pq+y19LqyY0L7+4Fy4IEq1rkwk8tDgPbA/ymltNb68zqUPxV4KnPBmDFj9gNe2bw5RDKZ3q6gs1FQkMeGDWUNVn5Tl/vrRHxOH0UF52PUUg+ZdTVkCLz9tpOzz/bzpz85uP/+KMOGJRsj5GajpV9bdSF1lT2pq+xJXWVP6ip71dWV2+1slMazpqBRE1Gl1F1Ab2CY1rqqjHAF8IK1br1S6iPgAOBza92uQMUAp27Ap5UL0FoXA8WVFnepnzMQ1XFGVuBb+yKRLudj5BTUef+BA9PMmhXmnHP8nHeen3HjYowbF5dBTEIIIcROrNHe5pVSt2OOeD9Rax2rZrPngWOs7YPAIGCute4VYIy1rjewPzCzIWMWWUpFyJt/OeAg0v3y7S6mQweDN98MM3Jkgrvu8jJ6tI/y8voLUwghhBBNS6Mkokqp/sB1QCfgK6XUHKXUG9a6OUqpTtamU4CuSqlfge+AZ7XWH1nrJgOtlVKLgXeBCysGNQkbpUK0mnManqJPKe87hbRvxxqfvV64994ot94aZeZMN8cdF2D5cnksqBBCCLEzaqxR878CVWYTFfOEWj9HgFHVbBfCnNpJNBGOZCmtZp+Cu/g7ygZMJ7bLyPop1wFjxiTo0yfNhRf6OfroANOnRznssFS9lC+EEEKIpkF64Int4kiU0Op/w3GX/EDpnk/VWxKa6fDDU3z4YYiCAoPTTvNzxx05JGUMkxBCCLHTkERU1J2RJu+XC3GX/UTpXs8R73Bigx2qRw+DmTPDjByZ5J57vJx8sp+1a+VWvRBCCLEzkERU1Flg2T14N35AeZ/biRcMbfDjBYNmv9EHHogwd66LIUMCfPKJq8GPK4QQQoiGJYmoqBNP0acEFk8k2vEUol3HNOqxTzstyUcfhSksNBg5MsDEiTkkEo0aghBCCCHqkSSiImvO6Cryfx5NKqgo63e/OaqokfXunWbmzDCjRsW57z4vJ54YYPVquVUvhBBCNEeSiIrspCLkzx0F6Tilez0LLvue+OD3w913x5g+PcK8eU6GDAny4Ydyq14IIYRobiQRFbUzUuT/cgHu0h8pG/AwqWBvuyMC4KSTkvznPyG6dEkzalSAG2/0Eo/bHZUQQgghsiWJqKiZYZCrr8G7/m1C6t/EC4fZHdE2evQweO+9MKNHx3n44RyGD5cJ8IUQQojmQhJRUSP/8nvxr3yE8K5jiXS7xO5wquTzwR13xHj88QiLF5u36l9+2Y1h2B2ZEEIIIWrSKE9WEs2Td+3L5C66kWiHkwn1vtXucGo1bFiSvfYKcemlPi67zM+sWQkmT47Spo3dkQkhhBANRynVB5gBtAOKgLO01osqbXMUcDuwB3C/1npcpfWnATdgPgnTAI7UWq9r6NilRVRUyREvInf+lcRb/4myAQ+Do3lcKt26Gbz5ZoR//SvG+++7OeywIP/9rwxkEkIIsVN7GHhQa90HeBCYXsU2vwHnA5Mrr1BK7QdMAP6itR4AHAKUNFi0GZpHdiEaXWD5vThS5ZT3nQJOr93h1InLBZdfHmfmzDB5eQannhrghhu8RKN2RyaEEELUL6VUIbAP8IK16AVgH6VUQeZ2WuvFWus5QFUPy74SuEtr/bu1bYnWutZ3TaXUG0qpE5VSnu2NXxJR8QfO2O/4V0wn1vFUUrl97Q5nu+25Z5qPPgpz3nlxpk/P4aijAvzyi1zyQgghmocPPvigo1Kqe6Wv1pU26wqs1lqnAKzva6zl2eoH9FBKfa6U+lEp9S+lVDYjf78AbgR+V0o9pJT6Ux2OCbSwPqJt2jT83JcFBXkNfowG98M/wYjj2/82fHkNdz6NVVePPWZ+gQuwb/7THbVTXFuNROoqe1JX2ZO6yp7UVfZqqqvJkye/UsXimzFvo9cnF7An8BcgB5gJrACermknrfU9wD1Kqf7A34AXlFJx4BngOa31ktoO3KIS0c2bQyST6QYrv6Agjw0byhqs/MbgjKyk7aLpRDuNojzaAaINcz521FVRkYNx47y8956HP/0pyQMPROnSpXkMrd8Zrq3GInWVPamr7EldZU/qKnvV1ZXb7aRNmyDjx48/9Yorrvih0uriSr+vBDorpVxa65RSygV0spZnawXwqtY6BsSUUm8BB1BLIlpBa/0rcJ1S6n3gAeAm4Gql1PfA1VrrudXt26ISUVG7wG93AhDu8Q+bI6l/7doZPPFElBdfTHL99T4GDw5yxx1RTj45acfTSoUQQogaDR069PehQ4cuq2kbrfV6pdQc4AzgWev7bK31hjoc6nngWKXUM5i54RHAq9nsqJRSmK2hZwIVraHHAxuAS4A3gd2q218SUbGFK7QY39rniHS5gLSvi93hNAiHA844I8nBB4e49FI/l1xiTvM0aVKU1pV73QghhBDNw0XADKXUjcBm4CwAq4XyRq31D0qpQ4AXgXzAoZQaCZyntf7QWr4fMA9IAx8Cj9d2UKXUD0B34CXgTK31t5U2uUcpNbamMhxGy5j1+xDgC7k1Xz3Pps/JmzcWZ2wdRYf8hOEtbNDjNYW6Sibh/vtzmDw5h8JCg/vvjzJoUMrWmKrTFOqruZC6yp7UVfakrrIndZW92m7NA4OALxs7rmwppU4B3tZab/cDtmUIcQvnSBSTO28srf93PAAl+7zW4EloU+F2w5VXxnn//TCBgMHJJ5vPq4/F7I5MCCGEaBZKMVtEt1Cmv2RbQNaJqFLqKqXUQOvng5RSK5RSS5VSB2dbhmhaXGW/0OarA/CteZZw9yvYdPA3JNr82e6wGt3AgWk+/jjMOeeYz6sfOjTAokXyGU0IIYSoxYNA5SbdMmt5VurybnslsNT6+d/APcBEYGodyhBNRSpK/s+jASg+4FNCvW8Bl9/moOwTCMCkSTGefjrMmjUOjjwywNNPe+R59UIIIUT1CrXWaystWwt0zLaAuiSirbTWJUqpPGAvzOeUPg6oOpQhmojg4ltwhxZQ1v9BkvkD7Q6nyTjmmBSffRZm//1TjBvnY9QoP2vXypB6IYQQogq/KaWGVFo2mK0Nl7Wqy6j5ldaM+f2Bz625qvKBpjm6Q1TLs+kL/CseJNLlPBLts+7G0WJ07Gjw8ssRHn3Uw+23exk0KMjNN8c488yETPMkhBBCbDUBeF0p9TiwBOgJnGt9ZaUuLaLjMeeU+idwq7XseOC7OpQhbOZIlJD360WkAj0o7zPR7nCaLKcTxoxJ8OmnIQYMSHHllT5OPdXPihWSiQohhBAAWuu3gKMwH1t4nPX9aGt5VrJuEdVav485U3+mV6wv0Uzk6mtwRldTvP8scDXfx102lh49DF5/PcKMGR5uucXLoYcGueGGGOeem8Ap45mEEEK0cFrr79iBRsmsE1GlVD+gSGu9TimVi9lCmgYmA4ntDUA0Hs/mr/CtfZ7QbuNItj7A7nCaDacTzj03wZFHJhk3zsd11/l46y03U6dG6dFDRjMJIYRouawZlQYB7YEttw211jdms39d2nReACqePXMXcChwEDC9DmUIuxgGgcW3kMrpQHi3cXZH0yx17Wrw4osR7rsvwvz5LgYPDjJtmoeU9JIWQgjRAimlLgT+DxgCXAPsAVwN9Mq2jLokot211lop5QBGAKcCpwBH16EMYRPPpk/IKf6KcI/x4ArYHU6z5XDAyJFJvvgixODBSSZM8HHccQEWLJD79EIIIVqcfwDHaK1PAiLW91Oow53yurx7Rq2pmw4AVmitNwIxwFeHMoQdDIPg4ltI+boR7Xy23dHsFDp2NJgxI8r06RGWLzfnHb377hx5KpMQQoiWpFBr/YX1c1op5dRafwAMy7aAuiSizwOfADOAp6xl+1CHuaKEPXI2vIendDbhHteA02t3ODsNhwNOOinJF1+EOfbYJHfe6WXIkABff+2yOzQhhBCiMaxSSnW3fl4InKCUGgRk/ez5rBNRrfWVmFM3Xay1fsBanMZ84pJoqow0wSUTSQZ6Ed3lDLuj2Sm1b2/wyCNRXnghTCzm4IQTAvz97z42bbI7MiGEEKJBTQL6Wj/fAjyL2Wh5c7YF1Kljm9Z6FrBEKXWwUqqb1voHrfUndSlDNC7v76/hLp9HuOf14KzL8wtEXR1xRIrPPw8xdmyMV15x8+c/B3nxRbc8JlQIIcROxxoz9DnwEYB1S74N0EZr/VC25WSdiCqldlFK/RdYBLwOLFZK/VcpVXluUdFEOBKbCS6+hWRuf2IdRtgdTosQCMANN8T5+OMwPXoYXH65nxEj/CxaJIOZhBBC7Dy01gbwM+bd8Yplca11eV3Kqcu740PAXKCt1noXzKx3DvBwXQ4oGomRJu+XMThjqynrOxUckgg1pn790rzzTpi77oryyy8uDj88wO235xAK2R2ZEEIIUW9mA312pIC63Ks9BNhFa50A0FqHlFL/AFbvSACiYQSW3o1340zK1CSSrQ+0O5wWyemEs85KcMwxSSZM8DJ1qpeXX/YwYUKME05IynPrhRBCNHefATOVUk8BK4EtndG01k9kU0Bdmsk2A/0qLVNAcR3KEI3AU/QJgSUTiXY8hWjXMXaH0+IVFhpMmxblnXfCtG1rcOGF5u36efOklVoIIUSz9mfM2ZMOA/4GjLK+/pZtAXVpEZ0EfKyUehxYDuwKnAvcUIcyRANzRlaS//NoUsHdKet3P9Ls1nQceGCKjz4K88wzHv79by9HHBFg9OgE//hHjFat7I5OCCGEqBut9eE7WkbWiajW+lGl1BLgTGBPYA1wptb6PzsahKg/wSW34UjHKN3rWXAF7Q5HVOJywTnnJBg+PMEdd3h5/HEPb7zh5p//jHPGGQmc0kgqhBCimVBKVfuupbVOV7cuU53m87GmapLpmpowz+YviLf7C6lgb7tDETVo2xYmTfr/9u47Tqry7P/4Z9pWVimiSBfFG0SqGNEoYEGFgMZYgvqDJ4oRiDG2YImimFgw2BUfeTQGlQRbVBQ7oKISjDQL6qVGBcSGisr2nfL74xxkXXdhVnf2zO5+36/XvHbmPmfPXHM7e7y4awXjxlVxwQW5nHVWHrNnx7jssgr22Ueb14uISJMQp9q40BrS2t1lq4moc+7P6VzEzC5O5zzJrHDZOiLl6yjr9vugQ5E09e2b5NFHy3jwwSh//nMuY8YU8KtfVXHxxRV07KgFSEVEJKvtUuP1zsD5wKPpXmBbLaJd6huRBCf29RIAKtv8POBIpD5CITj66DiHHx7npptymDkzhyefjFHQFfQAACAASURBVPKHP1QyeXIl+flBRygiIvJDZramRtEa59z/AK8Af0vnGltNRM3spB8ZmwQgtnEJyeh2JFr1CToU+REKC+H8872xopdemsv06bn8858xLrmkgpP0lygiIk3DdkD7dE9Oe4yoc65HHYcqgE/SHZQqmRP7eglV2+8DobSGZUiW6tYtxR13lPPCC1VcdFEuEybkM2cOXHJJmD320J+ZiIhkB+fc3Xx/jGgBMBRvz/m01GeO7nt423u+W+P5WqDCOfcv59xO9bieNKBQ5ZdES4yqNvsFHYo0kAMOSLBwYSnTp5fz6qtw0EEFnHdeLl99FXRkIiIigJcP/rfaYyneikqnp3uB+sya/y0wHJiGt3p+V+Ai4N/A88BVwEzgmHpcUxpI7Ot/A1DVWolocxKNwsknV/Hb3+YxZUoVs2fHePjhGOeeW8H//E8V0XqteyEiItJwzOzSn3qN+vxv7FJgNzMr91+/55z7HfCOmc1yzv0Gr4X0B5xz7YC7gV2BSv+8iWa2ocZ5s4FDgC/8ovvN7HL/2HN4ye+3/rEbzOzv9Yi/WYttfIlUOJf49oOCDkUyoG1bmD69gvHjq5g6NZcLLsjjrru85Z4OOEDLPYmISONzzt0I3GNmS6qV7QccZ2ZnpnON+nTNh4HuNcq6smWdqBLqTmxTwF/NzJlZX7zm2+l1nDvdzAb4j8trHPtDtWNKQquJfb2Equ0GQzg36FAkg/bYI8kDD5Rxxx1llJSEOProAk46KY81a7SDloiINLrjgWU1ypbjbX6Ulvq0iF4PLHLO/R2va74z3haf1/vHR+F10/+AmX0FPFetaCkwuR7vnTbnXGugdfWyWbNmdRg+fHgm3i47xIuJbnqN0u5nBR2JNIJQCEaPjnPwwXFuvTWHG27IYcGCKKedVsnpp1dSqA21RESkcaT4YaNmpJayOoVSqfQXzXbOHQ4cC3QEPgHuM7Mn074A320H9TTwiJndWOPYbLzZViV4raYXmNlb/rHngJ3wVvF/FTjPzNbXcv1pwCXVywYNGsTcuXPrE2bT8skz8OyhMPxJ6HhY0NFII/voIzjvPPjnP6FTJ5g+HU44AW0XKiLS9B0AvBh0EHVxzv0L+AA418ySfo43HehpZkelc436bvH5JFCvxLMWNwHFwM21HLsQfyko59x44EnnXA8zSwDjzGydcy4CXADcC+xfyzWuB2ZXL5g4ceJg4P6NG0uIxzO3/E379kVs2LApY9evS8GHz1BAmC9DfUkF8P4/RlB11VRtrb5yc+H66+H44yNcdFEu48ZFuOqqBFOnVnDggS1v/Ki+W+lTXaVPdZU+1VX66qqraDRMmzZNonvrDGA+8Ilzbg3ekM1PgDHpXiDtFlHnXAxvlvw4vBbRj/EmIF1uZpVpXuNqoB8wxswq0jj/S2BQzZX7nXNFwEYgJ831S/cHXmiuiej2y0YRipfw9ZDnG/29fyzdqOon3fpKJuGhh6JceWUua9eGGTo0ztSpFfTv33LWH9V3K32qq/SprtKnukpfGoloVreIwnc93T/D241zHfCf+qwtX5/Ou7/izWifBPT3fx6Et2xTOoFeAewF/LKuJNQ516na88OABLDeORetsUbp8cDrWkQfSFYQ+2YZVW32DToSyQLhsLdd6EsvlXDZZeW88UaYESMKmTQpjw8/1IQmERFpOM65AUAnM1tqZveb2VKgk3Ouf7rXqE/X/LFAfzP70n9tzrkVeOM1tzpLxjnXB687/R1giXMO4AMzO8o5twoYZWYfA3f6CWcSb5mmI8ws7pwrBB5zzuUAIWA9MLYesTdb0W9XEUqWa/1Q+Z7cXDj11CrGjq1i5swcbr01h0cfjfKb31Rx1lmV7LBD+mPDRURE6jAHOKJGWQ5ej3m/dC5Qn0S0ruaUbTazmNnqus4zswHVnh9SxzklwOA0Ymxxcj9/lFQoSlWbnwcdimSh7baDCy6o5KSTqpgxI4c77ogxd26M00+v5NRTNcNeRER+kq5m9n71AjP7r3Oue7oXqE/X/P3Ao865w5xzvf0Z9A/75RKEZCV5n8ylsv1IUjntgo5GsliHDimuuaaCxYtLOeCAOFdemcuQIYXceWeMyrRGeIuIiPzAR8657+2k47/+ON0L1CcRPRdYgLeN53K82e/PAlPqcQ1pQDkbniRcuYHyjuOCDkWaiJ49k9x5ZzmPPlpKt25JpkzJY7/9CpkzJ0ZVVdDRiYhIE3MdMM85d7pzbpRz7nTgIeDadC+w1a5559xBNYqe8x8hvEVMwZuRvijdN5SGk/fxXSRyO1LZrtYRDSJ12mefBI8+WsaiRRH++tdczj47j+uvz+Hssys49tg4sVjQEYqISLYzs9ucc18DE/Bmza8FzjGzB9K9xrbGiP6tjvLNSejmhLRHum8oDSNcvp6cLxZQusvZEK7XcrAigLdD08EHJzjooFIWLIgwY0YuZ56Zz3XXJTnnnAqOOSZOVF8tERHZusVABbCD/3o759zJZnZHOr+81f/NmNkuPzE4yZC8j/9BiKS65eUnC4VgxIgEhxxSytNPewnpH/6Qz7XXJjnrLC8hVQupiIjU5Jz7Jd4M+feAPsBqYE+8tU/TSkS1CWBTlEqSt/5uKtsOI1mgfytIwwiF4LDDEjzzTCl33VVKUVGKM87IZ999vUlNFdvcgkJERILgnNvdOfdv59w7/s+etZxzqHNumXOuwt9gqLbrOOdcaV3Ha3EZcLKZDQRK/J+n4s0lSosS0SYo9tViIuVrKO84PuhQpBkKheDwwxMsWFDKP/5RSvv2KaZMyeNnPyvktttilJUFHaGIiNRwKzDTzHbHm1Q+q5Zz3gdOAWbUdgF/C/VZeCsipaurmdVcPelOIO0ERYloE5S3/k6S0dZU7Jj2Vq4i9ba5y/7xx0u57z5vlv2FF+YxeHAhM2fGKC4OOkIRkebtiSee6OCc617j0br6Oc65HYFBwFy/aC4wyDnXvvp5Zvaema0C4nW83fl4+8a/U48QP6+28+WHzrl9gV2BSLoXaFFTEfx9WzOqffuizL5B+eewYT7sdirtO7Tf9vlZLON11cwEWV/HHus9PGEgz39kJ3230qe6Sp/qKn2qq/Rtra5mzJhR21rtlwLTqr3uAqw3swSAmSWccx/75RvSicHfkvMw4EBgalqBe27DWz3pX3hLOT2LtzvmNeleoEUlohs3lhCPZ257+vbti9iwYVPGrg/Q6s0LyEsl2LjDb0hk+L0yqTHqqjnJtvpatizMjTfm8OSTMfLzUxx/fBWTJ1fSrVvwW4dmW11lM9VV+lRX6VNdpa+uuopGw7RpU8iUKVOOPfPMM5fVOPx1Q8bgnIsB/wec5Cexaf+umV1V7fldzrnngEIzeyvda7SoRLSpi2xaTd762ZR1OZVE4Q/GIYs0msGDk9x1VznvvFPJzJk53H13jDvvjHHkkXFOO62Svn0z9w8+EZGWYuTIkZ+OHDnyw22ctg7o5JyL+IlkBOjol6djZ7zu9Mf9JLQ1EHLObWdmp9YnXjNbW5/zQWNEm45Uilbv/IlUdDtKdz0/6GhEANh99yQ33FDOsmUlTJxYxdNPRzn44EKOOy6fxYsjpIJvIBURadbM7HNgFXC8X3Q8sNLM0uqWN7O1ZraDmXU3s+7A9cBt9U1Cfywlok1EzhdPk/PVs5T2OI9UrG3Q4Yh8z847p5g2rYKVK4u56KIKVq8Oc8wxBRx6aAHz5kVJJIKOUESkWZsEnO6cewc43X+Nc+5x59xg//n+zrmPgLOBic65j5xzhwUWsS+UahlNFvsDLzTZMaLJKtos3RdSCTbu+zKEcxr+PRqZxhDVT1Orr/JyuP/+GDNn5vD++2G6dUvyu99VMnZsFfn5mX3vplZXQVJdpU91lT7VVfq2NUYUOABvcfhmSy2iTUDeR3cQLXmHkp6XN4skVJq/vDwYN66Kl14q4Y47ymjXLsV55+Wx116FXHttDhs3Bh2hiIhkAyWi2S5RRuH7V1HZZiiV7UcGHY1IvUQiMHp0nCeeKOXhh0vp3z/J9Om5DBzYiqlTc1m/PhR0iCIiEiAlolku75N7CVd9QWmP87wVxkWaoFAI9tsvwdy5ZTz7bAmjRsW5/fYYe+9dyO9/n8dbb+lWJCLSEunun81SKfLXzqSqqD9VbfYPOhqRBtGnT5JbbinnP/8p4eSTq5g/P8qwYYWceGI+//63ZtqLiLQkSkSzWOzLBURLjLJup6k1VJqdLl1SXHZZBStWFHPeeRWsWBHmyCMLGDWqgEcfjRKvaxM6ERFpNpSIZrGCNTeTyOlAxU6/CjoUkYxp2xbOOaeS5ctLmD69nC++CDFhQj777FPILbfE+OaboCMUEZFMUSKapSKbVpPz1bOUdZ2omfLSIhQUwMknV7F0aQmzZ5fRpUuSadPy6N+/FVOm5GocqYhIM6Q7e5bKX3sLqXAB5Z1PCjoUkUYVicCoUXEefriMhQtLOOKIOPfcE2PYsEJ++ct8HnkkSlVV0FGKiEhDUCKahUIVn5P3yb2UdzxBuyhJi9a3b5Ibbyxn1aoSpk6t4KOPwpxySj577VXIjBk5fPaZxk6LiDRlSkSzUMHa/4VUFWVdJwcdikhWaNcuxemnV/LyyyXMmVPKHnskmTEjl4EDCzn11DyWLtVsexGRpkiJaJaJfvsq+WtuoKLDMSQKewYdjkhWiUTg0EMT3HNPGUuXFjNhQhXPPhvliCMKOOigAubMiVFaGnSUIiKSLiWi2SRRTtEbp5KM7UCx+2vQ0YhktR49UvzlLxWsWlXMNdeUk0rB2Wfn0akTXHxxLh98oG57EZFsp0Q0ixT+9y9ES96iuM/NpHLaBR2OSJNQWOjta//ss6U88kgphx0Gt98eY8iQQk44IZ+FCyMkk0FHKSIitVEimiViX71I/pqbKes8gcodDg06HJEmJxSCIUMS3HMPrFhRwjnnVPLaa2GOP76AIUMKufXWGF9/HXSUIiJSnRLRLBCKf0vR6kkk8nehePfLgg5HpMnr0CHFuedWsmJFCbNmlbHjjkkuvjiPAQNacc45WpNURCRb6G6cBXI/nkukfC2b9rwVIoVBhyPSbOTkwFFHxZk/31uT9Kijqrj/fm9N0qOOymf+fG0lKiISJCWiWSC66TWSsR2Ib79P0KGINFt9+ya57jpvctPUqRWsWRPm5JPz+dnPCrnxxhy+/FKTm0REGpsS0SwQLV5NvKiPN8hNRDKqbVs4/fRKXnnF20q0e/ckl13mrUl6xhl5vP66bosiIo1Fd9ygpRJEi98i3qpP0JGItCibtxJ98MEynn++hOOOq2LevCgHH1zI6NH5PPywthIVEck0JaIBi5R+QChZRkKJqEhgevdOcvXVXrf9pZeW89lnYU491eu2nzkzxrffBh2hiEjzpEQ0YJHi1QBqERXJAq1bw+TJVSxdWsLdd5fSvXuSSy/No3//VkydmsvatRo+IyLSkJSIBixavJoUIeKtegUdioj4IhE47LAEDz1UxoIFJRx+eJy//S3Gz35WyCmn5LFsmW6dIiINQXfTgEWLV5Mo2BUiBUGHIiK16Ncvyf/+bznLlpVw2mmVPP98lFGjChk5soB//CNGcXHQEYqINF1KRAMW2bRa40NFmoCOHVNMnVrJypXFXHFFOcXFcNZZefTt6y2Sv2JFmFQq6ChFRJoWJaJBSpQQKftA40NFmpBWreCUU6pYvLiUxx4r4Ygj4vzrXzEOP7yQ4cMLuO22GBs3Bh2liEjToEQ0QNHitwiR8tYQFZEmJRSCvfdOcsMN5bz+ejFXX11Obi5ceGEe/fq1YtKkPF58MUIyGXSkIiLZS4logKKbNGNepDkoKoLx46t4+ulSFi4s4cQTq1iwIMqvflXAkCGF3HBDDuvXa8a9iEhNSkQDFCleTSpSSDK/e9ChiEgD6ds3yfTpFbz+ejEzZ5ax885JLr88l0GDvP3t//EPrUsqIrKZEtEARYtXE2/VG0L6zyDS3OTnw7HHxpk3r4yXXy5mypRKPvkkzFln5dGnTytOOSWPJ5+MUFkZdKQiIsFRBhSUVMpPRNUtL9Lc7bJLij/+sZJ//7uEJ58sYdy4Kl56KcL48QX061fIeefl8sormnUvIi1PtDHexDnXDrgb2BWoBN4FJprZhhrnzQYOAb7wi+43s8v9Yzv51+gOlAGnmtnLjRF/JoQrPiVc9ZUSUZEWJBSCQYOSDBpUwaWXVvDccxEeeCDG3Lkx/v73HLp3T3LMMVUcdVScnj01y0lEmr/GahFNAX81M2dmfYH/AtPrOHe6mQ3wH5dXK78SWGxmuwOnAXOcc0129P/mrT21hqhIyxSLwYgRCWbNKmf16mJuvLGMLl2SXHNNDj//ubcU1HXX5fD++032Nicisk2N0iJqZl8Bz1UrWgpMrudljsNrDcXMXnTOVQCDgVeqn+Scaw20rl42a9asDsOHD6/n22VW9Ls95vcIOBIRCVpREYwdG2fs2Diffhpi/vwoDz8c5corc7nyylz69k1w5JFxxoypYpdd1H8vIs1HKNXIg5Kcc2HgaeARM7uxxrHZwFCgBK/V9AIze8vv2l9rZoXVzn0cuN3MHqxxjWnAJdXLBg0axNy5czPwaX6CJePhs4Vw1PqgIxGRLPXRR/DAA3DvvbB0qVe2117w61/DscdC9+6BhicimXcA8GLQQWRSo7SI1nATUAzcXMuxC4FPzCzpnBsPPOmc61HP618PzK5eMHHixMHA/Rs3lhCPZ27cVfv2RWzYsCmtc9t8sYpkwR58k+b5zU196kpUX/XRnOoqNxdOPNF7rFsX4tFHo8ybF+PccyOcey4MGpTgiCOqOOKIOJ07179RoTnVVaaprtKnukpfXXUVjYZp06awlt9ofho1EXXOXQ30BMaY2Q8yQjNbX+35Xc6564DOZrbGOYdzbgcz2zyRqSuwrpZrfA18XaO4c4N9iIaQrCJSYlS2OyjoSESkiejSJcXvflfF735XxZo1IR55JMYjj0SZNi2PadNg8OAERx5ZxZgxcTp2VPe9iDQNjbZ8k3PuCmAv4JdmVlHHOZ2qPT8MSACbk9P7gUn+sf2BfGB5JmPOlEjpu4RSlRofKiI/SrduKU4/vZJnnill6dJiLrywgvJymDo1jwEDWjFmTD633x7j00810UlEslujJKLOuT7ABUBHYIlzbpVz7iH/2CrnXEf/1Dudc687514FLgKOMLO4f+x8YLhz7l3gFmBcba2qTUH025UAxLcbGHAkItLU9eiR4owzKlm0qJQlS4o5//wKvv02xJ/+lEf//oUceWQ+d9wR4/PPlZSKSPZp9MlKAdkfeCFbxogWvj2F/I/n8MWBH0EokrF4spnGENWP6it9qiuPWZhHHokyb16Ud96JEA6n2HffBKNHx/nFL+J06JBSXdWD6ip9qqv0pTFGtNlPVtLOSgGIfbuKeFHfFpuEikjmOZdkypRKXnihlOefL+HMMyv5/PMQF1yQR79+rRg1qoBrroG1a9VSKiLBCWLWfMuWjBPd9BplnX8TdCQi0gKEQtC7d5LevSs5//xKzMLMnx9l/vwof/wjQCv69fNaSkePrmK33VpEL5lIs+Kc2x24E2gHfAmMN7N3a5xzKHAF0Be4ycz+WO3YVGAs3tycKuBPZvZUY8SuFtFGFil9h1CyjHjRgKBDEZEWyLkk55xTybPPlvLeezB1agXRKFxxRS777deKoUMLuOqqHFavDtMyRm6JNAu3AjP93SdnArNqOed94BRgRi3H/gPsbWb9gJOBe51z+ZkKtjoloo1sy0SlQQFHIiIt3a67wumnV/Lkk6WsXFnM5ZeX06ZNimuvzeHAAwsZMqSQv/wlh5UrlZSKBOGJJ57o4JzrXuPxvd0jnXM7AoOAzTv3zAUGOefaVz/PzN4zs1VAnBrM7CkzK/VfvgaE8FpXM65Fdc03xuKw7dsXbf2ENW9CtJC23QZCuGWPEd1mXcn3qL7Sp7pK3+a6at8eBgyAP/2p+tEwkOs/RN+r9Kmu0re1upoxY8b9tRRfCkyr9roLsN7MEgBmlnDOfeyXb/gRIY0H/mtmH/2I3623FpWIZsOs+dafvQyt+vH1l6VbPa+506zK+lF9pU91lb506mrjRnjqqSjz58d47rkIlZUhdtwxyahRcUaPjrPffgmiLeD/JPpepU91lb5tzZqfMmXKsWeeeeayGodrbtrTYJxzw4C/ACMy9R41tYDbRxZJxoluep2yzicFHYmISFratIGxY+OMHRtn0yZ45hlvotN998WYPTuHtm2THH64tyTU0KEJctV4KtJgRo4c+enIkSM/3MZp64BOzrmI3xoawVu3/Qe7T26Nc25fYA5wpJnZjwr4R9AY0Ub03USl7TRRSUSanqIi+NWv4txxRzlvvlnMHXeUMXx4gkcfjXHiiQXssUcrJk3KY/78KKUtu9NHpNGY2efAKuB4v+h4YKWZpd0t75zbG7gXOMbMVjR8lHVTi2gj+m6iUpF2VBKRpq2gAH/JpzgVFbB4cYTHHovyxBMxHnwwRkFBioMO8o6PGBGnSEMGRTJpEt7ulBcDG/HGeeKcexy42MyW+duj3wNsB4Scc2OBCf4yTbfgbZ0+yzm3+ZrjzOz1TAeuRLQRxb5dSTLSikThbkGHIiLSYHJzYcSIBCNGJLj66gqWLIkwf36Uxx/3xpbm5KQYPjzB6NFVHHZYnDZtgo5YpHkxs7eBfWopH1Xt+YtA5zp+f+/MRbd1SkQbUfTbVcSL+mlHJRFptqJRGDo0wdChCa68soJXXvFaSufPj/L00/lEo95WoyNHxjn00Dhdu2pdKJGWTGNEG4s/UUnjQ0WkpYhEYMiQBH/5SwUrVpTw1FMlTJ5cyaefhvjTn/IYPLgVw4YVcOmluSxaFNG4UpEWSC2ijSRSYv5EJY0PFZGWJxSCgQOTDBxYydSplbz/foinnoryzDNRbrstxsyZOeTkpBg82GtNPeCAOAMHJlvE0lAiLZn+xBtJdNMqACWiIiJAjx4pJk+uYvLkKkpL4eWXIyxeHOWFFyJcdVUO06fnUlSUYr/9vKR06NAEziUJhYKOXEQakhLRRhL7doU3UalAE5VERKorKIADD0xw4IEJAL76Cl56Kcrzz0d44YUoTz2VB8COOyY54IAEQ4fGOeCABJ07a3ypSFOnRLSReBOV+kNIw3JFRLambVsYMybOmDFxoIJ160K88ILXYvr88xH+9a8YAD16JPn5z+MMGZJgyJAEXbooMRVpapSINoZUimjxm5R1HBd0JCIiTU6XLilOOCHOCSfESaXg7bfDLF7stZbOmxfj7rtzAOjcOcmBB8Y55BCvO79Vq4ADF5FtUiLaCEJVXxJKlJAs6B50KCIiTVooBL17J+ndO8nEiVUkEvDWW2FefjnCiy9GeOghLzHNyfGWiRoxIs4hh8Tp0UOtpSLZSIloI4iUrQEgkdc92EBERJqZSAT23DPJnnsmmTChispK+M9/IjzzTJSFCyNcdFEeF13kdeMfcoiXlO67b4Lc3KAjFxFQItoovktE87sGHImISPOWkwP7759g//0TXHopfPhhiIULoyxYEOWuu2L83//lUFCQYuhQrwv/kEPidOyo1lKRoCgRbQTh8rUAJJWIiog0qu7dU0yYUMWECd4yUS+95LWWLlgQ5cknvUlPe+zhdeEffHCCwYMTWrtUpBHpz60RRMrWkIy1IRXdLuhQRERarIICGDEiwYgRCVKpCszCLFgQYcGCKDffnMMNN4Ro3TrlT3iKM3x4gvbt1VoqkklKRBtBpGyNxoeKiGSRUAh69UrSq1eS3/++im++geef91pKFyzwJj2FQin6909y0EFxjj4aevTwxqSKSMNRItoIwmVrSLTaI+gwRESkDttvD0ccEeeII+Ikk/Daa2EWLYqycGGU66/P4dproXXrVgwbFufgg+MceGCCnXZSa6nIT6VENNNSKSLl66hsPzLoSEREJA3hMAwYkGTAgErOPruSr7+GlSuLeOihOIsWRZg3zxtbuueeCQ4+OM5BB3ljS2OxgAMXaYKUiGZYuPIzQslyzZgXEWmiWreG446DAw8sJ5WCN94I8+yz3vJQm8eWFhV5M/EPPjjBgQfG6dRJraUi6VAimmFhf+mmZF63gCMREZGfKhSCvn2T9O1byR/+AN9+C4sXR3n22QgLF0Z57DGvWXSXXbztR3/+8wT77Zdg552VmIrURolohm1ZQ1SJqIhIc7PddjB6dJzRo+PfzcR/7rkIS5ZEeOSRGHPmeNuP9ujhJab77afEVKQ6JaIZFvHXEFXXvIhI81Z9Jv6kSd72o2++GebFFyMsWRJl3jxv+1GArl2TDBmSYMiQBPvu621BGgoF/AFEAqBENMPCZWtI5rSHSEHQoYiISCOKRDZ34yeZPNlLTN94I8zSpRGWLo2waFGE++7zuvJ32GFLYjpkSII+fZJaKkpaBCWiGRYpW6tueRERIRKB/v2T9O+fZOLEKlIpeO+9LYnp0qUR5s/3EtOiohR7770lMR0wIEFeXsAfQCQDlIhmWKTsQ6q2Gxh0GCIikmVCIejZM0nPnknGjasCYP360PcS0yuuyAUgNzfFwIFbEtO9905QVBRk9CINQ4loJqUShMs/IrnTUUFHIiIiTUCnTimOPjrO0UfHAfjyyxD/+c+WxPSmm3K4/voQ4XCKPfdM8rOfeUnp3nsn6NRJ40yl6VEimkHhik8Ipao0UUlERH6Udu1SjBwZZ+RILzEtLobly72k9OWXI/zznzFuv92bALXzzsnvktLBgxP07ZskJyfI6EW2TYloBkXKNs+Y1xhRERH56Vq1gmHDEgwblgAgHvdm5r/ySoRXXomwbJm3bBRAXl6K/v03t5gm2WuvBDvuqGWjJLsoEc2gcNmHACTz1CIqIiINLxqFfv2S9OuXZMIEb5zpp5+GvktMX3klwqxZOdx8s9dn37VrdFlRfgAAEOlJREFUkkGDEuy1V4JBg7xWU02CkiApEc2gLS2iSkRFRKRxdOiQYsyYOGPGeN355eXw6qsRli8Ps2KFl5w+/LDXahqLeWNNqyenu+yisabSeJSIZlCkfA2J3J0hnBt0KCIi0kLl5cE++yTYZ58EsKXVdPnyCCtWeMnp3Lkx/vY3b0Bp27ZJBg36fnK6/fYBfgBp1pSIZlC4bC1JjQ8VEZEs06FDil/8Is4vfuG9jsfBLPy95HThwhxSKa9ptGfPBIMGeeNM99orQe/eSaLKIKQB6GuUQZGyNVS1HhJ0GCIiIlsVjUKfPkn69EkyfrxXtmkTrFwZYcWKCMuXR1i4MMK993pd+vn53kSo6slpx46aCCX1p0Q0U5JxwhXrNWNeRESapKIiGDo0wdCh3gz9VArWrg2xYoWXnC5bFuH222PccovXpd+hg5eUDh0KvXpF6NcvQWFhkJ9AmgIlohkSrviIUCqhrnkREWkWQiHo1i1Ft25xjjrKmwhVUQGrV4e/azVdvjzCY48BFBCJpOjVa0uL6V57JdlttyThcKAfQ7KMEtEM0RqiIiLS3OXm4k9sSnLKKVV+aRHPPFPqjzf1ZujfdZfXarrddin23DPBHnt4wwD22COBc0kKCoL7DBIsJaIZEilbA0BCa4iKiEgL0r49jBiRYMQIr0s/mYT//jfM8uXeZKjVq70doUpLvYlQoVCKHj1S9OmToE+fJAMHJhgwIEHr1kF+CmksSkQzJFy2hhRhknmdgw5FREQkMOEw9OyZpGfPJGPHel36ySSsWRPizTcjrF4d5s03w7z22pZdoQB23dVLSgcNSjBwoJekavH95qdRElHnXDvgbmBXoBJ4F5hoZhvqOH84sBA4w8xu9sueA7oC3/qn3WBmf89s5D9etOQdknmdIBzb9skiIiItSDgMu+ySYpddtiwhBfDNN7BqVcSfrR9m8eIIDzywZfF955L+7P4Ee+7p/WzTJqAPIQ2isVpEU8Bfzew5AOfcDGA6MKHmic65IuAq4IlarvMHM5ufwTgbRPTbleR8/gjlXU4JOhQREZEmY/vtYdiwBMOGbZmp/8kn3kz9lSvDvPFGhEWLtiwjBdCxY/K7pHRzkrrLLilNimoiGiURNbOvgOeqFS0FJtdx+rXADGD0j3kv51xr4HsjS2bNmtVh+PDhP+Zy9ZdK0OrNM0nmtKdk16mN854iIiLNUCgEHTum6NgxzuhqWcHnn4dYvdpLTDd37S9cmEMi4Y07LShI0bv3luR0zz29RfhbtQrog0idQqlU4y5A65wLA08Dj5jZjTWOjQROMrPjnHOzgWU1uuZ3AuLAq8B5Zra+lutPAy6pXjZo0CDmzp3b8B+mNnYjLD8Dfn4PdPt147yniIhIC1deDqtXw6uveo9Vq7yf33zjHQ+FYLfdoH9/GDBgy89OnbxjWeoA4MWgg8ikICYr3QQUAzdXL/RbMqcDI+r4vXFmts45FwEuAO4F9q/lvOuB2dULJk6cOBi4f+PGEuLx5E+LfivaF35DctVFxNsdwjf5I2HDpoy9V1PXvn0RG1Q/aVN9pU91lT7VVfpUV+kLsq66dvUeY8Z4r1Mp+Oij0Hctp6tXh3nllQgPPLCl375NG2/Gfu/eSXr1StK7d4JevRqn9bSuuopGw7Rp0zJ2A2jURNQ5dzXQExhjZjUzwj2BnYH/OOcAdgDGOOfamtmfzWwdgJklnHM3ANOcc+Ga1zGzr4Gva1y7caauLz+DUKqKTb2uyep/XomIiLQEoRB06ZKiS5c4I0duKd+0CVav3pKcvvlmhH/8Y8uSUgBdu3qJaa9eW5LU3XZLkpsbwAfZBufc7sCdQDvgS2C8mb1b45xDgSuAvsBNZvbHasciwI3A4Xjzeqab2e2NEXujJaLOuSuAvYBfmFlFzeNm9iKwY7XzZ+N3zTvnokA7M/vMP3w88HotyWxgYhuXwLoHKdntEpIFuwQdjoiIiNShqAiGDEkwZEjiu7Jk0tvC9O23w7z9doS33grz9tthFi3KIR73EtRo1Bt7euedZXTu3LhDG7fhVmCmmc1xzv0/YBZwUI1z3gdOAY4Bai6EdSKwG15jYTtgpXNugZl9mNGoabzlm/rgdae/AyzxWzw/MLOjnHOrgFFm9vFWLpELPOacywFCwHpgbIbDrpdkzo7Q+1zKOp4edCgiIiJST+EwdO+eonv3BIcfviVBraz0FuT3EtQwn30WIicnwEBrcM7tCAxiy9DGucDNzrn21ZfJNLP3/PN/Wctlfg3c5jfwbXDOPQwcizd5PKMaa9b8arwEsrZjA+oo/0215yXA4IwE10AShbtB96s0LlRERKQZycmB3r2T9O7d+J2wTzzxRIczzzyze43ir/1hiJt1AdabWQK+G8L4sV9e63rttegKrKn2eq3/+xnXonZWaoyBv+3bF2X8PZoL1VX9qL7Sp7pKn+oqfaqr9Kmu0re1upoxY8b9tRRfCkzLVDyNrUUlohmfNa9ZlWlTXdWP6it9qqv0qa7Sp7pKn+oqfduaNT9lypRjzzzzzGU1DteckL0O6OSci/itoRGgo1+errVAN+AV/3XNFtKMaVGJqIiIiEhTMXLkyE9Hjhz54dbOMbPP/fk2xwNz/J8r69pGvQ73A791zj2IN1npl3hrmGacElERERGRpm0ScKdz7mJgIzAewDn3OHCxmS1zzu0P3ANsB4Scc2OBCWb2FHA3sA+wecmnP5vZB40RuBJRERERkSbMzN7GSyRrlo+q9vxF6lhX3Z/oVNfW6xkV3vYpIiIiIiINT4moiIiIiARCiaiIiIiIBEKJqIiIiIgEoqVMVooBRCKZz7ujUeX26VJd1Y/qK32qq/SprtKnukqf6ip9tdVVtXwl1qjBBCCUSqWCjqExTAZuCToIERERkXr4HfC/QQeRSS2lRXSO//NtoCrIQERERES2IQb0Ykv+0my1lBZREREREckyGsQhIiIiIoFQIioiIiIigVAiKiIiIiKBUCIqIiIiIoFQIioiIiIigVAiKiIiIiKBUCIqIiIiIoFoKQvaZ5xzbnfgTqAd8CUw3szeDTaq4Dnn2gF3A7sClcC7wEQz2+CcGwLMAvKBD4H/Z2afBxVrNnHOXQJMA/qa2Ruqq9o55/KA64BDgHLg32Z2qv4ef8g5Nxr4CxDyH5ea2YOqK3DOXQ0cDXTH/5vzy+usm5Zab7XV1dbu8/7vtMj7V13fq2rHv3ef98taXF2pRbTh3ArMNLPdgZl4XySBFPBXM3Nm1hf4LzDdORfG2zHiNL/OFgPTA4wzazjnBgFDgDX+a9VV3f6Kl4Du7n+/pvrl+nusxjkXwksUxpnZAGAccKf/3VJdwcPAUPy/uWq2Vjcttd5qq6ta7/PQ4u9fdX2vfnCf98taZF0pEW0AzrkdgUHAXL9oLjDIOdc+uKiyg5l9ZWbPVStaCnQD9gLKzexFv/xW4LhGDi/rOOdy8f6nNrlaseqqFs65VsB4YKqZpQDM7DP9PdYpCWzvP28NfALsgOoKM3vRzNZVL9va96glf8dqq6ut3OehBd+/aqsrqPM+Dy20rpSINowuwHozSwD4Pz/2y8Xn/2tvMvAI0JVq/xI0sy+AsHOubUDhZYs/A3PM7MNqZaqr2u2K1yV6iXNumXPuOefc/ujv8Qf8RP04YJ5zbg1eS814VFdbs7W6Ub3VocZ9HnT/qk1t93looXWlRFQa001AMXBz0IFkI+fcvsBg4JagY2kiIkAPYKWZDQbOAx4EWgUaVRZyzkWBC4AjzawbMAa4D9WVNDzd57dC9/kfUiLaMNYBnZxzEQD/Z0e/XPhu0HZP4NdmlgTWsqXrBufcDkDSzL4KKMRsMAzoDXzgnPsQ6Aw8BeyG6qo2a4E4fveomb0MfAGUob/HmgYAHc3sJQD/Zwne+FrVVe22dl/XPb8WtdznQff6mmq9zzvnDqWF1pUS0Qbgz2hbBRzvFx2P10qzIbiosodz7gq8sS+/NLMKv3g5kO93pQJMAu4PIr5sYWbTzayjmXU3s+7AR8BhwAxUVz/gd1s9C4yA72Yx7wi8g/4ea/oI6OyccwDOud7ATnizm1VXtdjafV33/B+q4z4Putd/T133eTN7mhZaV6FUKhV0DM2Cc64X3lIebYCNeEt5WLBRBc851wd4Ay85KPOLPzCzo5xz++HNNM1jyzIVnwUSaBby/7U82l8eRXVVC+dcD+AOvCV0qoALzewJ/T3+kHPuROB8vElLAJeY2cOqK3DO3Qj8CuiA16r+pZn12VrdtNR6q62u8MYf13qf93+nRd6/6vpe1TjnQ/z7vP+6xdWVElERERERCYS65kVEREQkEEpERURERCQQSkRFREREJBBKREVEREQkEEpERURERCQQSkRFJKs45z50zh0ScAy3OuemBhlDJjjnfuOce3HbZ4qINI5o0AGIiATJOfcb4BQz27yINGY2KbiIRERaDrWIiogExN8DXkSkxdJNUESy0d7+riQ7Aw8Dk82sHMA591vgPKAt8CIwycw+9o/tB9wA7I63y8sZZrbEP/Yb4GKgPd4uJxcBK4BbgZhzrhiIm1lr59xs4CMzu8g5NxyYA1znv28C+JOZ/d2/bjtgNt4e0gY8BQyv3sK6mXOuO/ABcApwCd7OKUOdc0cAVwKd8LaOnGxmb/m/kwJ6mtl7/uv6xvZ3YDjwth+biEjWUIuoiGSjE4HDgF3xksqLAJxzB+ElbMfhJalrgHv8Y22Bx4Ab8bb8vBZ4zDnXzjlX6JePNLMiYD9glZ/sTQL+bWatzKx1HfF0ALbHSxQnADOdc238YzOBEv+c//Ef2zIM6A0c5pzbHZgLnImXJD8OPOqcy0njOunEVo5XVyf7DxGRrKFEVESy0c1mts7MvgIuB473y08E7jCzFWZWAVwA7Ou3NP4CeNfM7jazuJnNxWsFHOP/bhLY0zmXb2afmNnqesRTBfzZzKrM7HGgGHDOuQhwNN6+7aVm9ibe/uPbMs3MSsysDPg18JiZPWNmVcDVQD5estwQsV3sv9cbacYmItJolIiKSDZaV+35GqCj/7yj/xoAMysGvsRrDfzesWq/28nMSvASvknAJ865x5xzveoRz5dmFq/2uhRohdeCGa0Rb/Xndal+Ts3PlPSPd8pAbDXrR0QkUEpERSQbdan2vCvwsf/8Y6Db5gN+l3s7YH3NY9V+dz2AmT1lZiPwuqnfBm7zz0n9hDg3AHGgcx2x16X6e9b8TCH/Guv9olKgoNr5HeoZW826FBHJGpqsJCLZ6DTn3Hy8JOxC4F6/fC4w1zn3T+At4ArgZTP70Dm3CbjJOXcCcB9et/QewHzn3E7AEGABUIbXfZ30r/kZ0Nk5l2NmlfUJ0swSzrkHgWnOuVPwEr3xwNp6XOY+4Hzn3MHAYuAMoAJY4h9fBZzgnFsNjMAbX7qsnrGdDHTHG7/6YT1iExHJKLWIikg2+ifwNPA+8F/gMgAzWwBMBf4FfII3mWmsf+xLYDRwDl53/bnAaDP7Au9edzZe6+NXeMncZP+9FgGrgU+dc1/8iFh/jzdZ6FPgbrxkuSLdXzYzA/4fcBPebP4xwJhqSfEZftnXeGNkH65nbK382GbjzaAXEckaoVTqp/RKiYhIdc65q4AOZpbO7HkRkRZNXfMiIj+BP+kpB3gd2BtvCaVTAg1KRKSJUCIqIvLTFOF1x3fEG296DTAv0IhERJoIdc2LiIiISCA0WUlEREREAqFEVEREREQCoURURERERAKhRFREREREAqFEVEREREQCoURURERERALx/wE4lRItX5J8hgAAAABJRU5ErkJggg==\n"},"metadata":{}}]},{"metadata":{"_kg_hide-input":true,"trusted":true},"cell_type":"code","source":"params['num_boost_round'] = len(result['multi_error-mean'])\nbst = lgb.train(params, dataset)\ns = pd.DataFrame({'feature': X_tr.columns,\n                  'gain': bst.feature_importance(importance_type='gain')})\ns.sort_values(by='gain', ascending=False).head()","execution_count":17,"outputs":[{"output_type":"execute_result","execution_count":17,"data":{"text/plain":"              feature          gain\n21   q95_roll_mean_64  31458.374495\n1                mean  23869.858433\n22   q05_roll_std_512  21763.000724\n18    q05_roll_std_64  21474.502374\n26  q05_roll_std_4096  20740.235864","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>feature</th>\n      <th>gain</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>21</th>\n      <td>q95_roll_mean_64</td>\n      <td>31458.374495</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>mean</td>\n      <td>23869.858433</td>\n    </tr>\n    <tr>\n      <th>22</th>\n      <td>q05_roll_std_512</td>\n      <td>21763.000724</td>\n    </tr>\n    <tr>\n      <th>18</th>\n      <td>q05_roll_std_64</td>\n      <td>21474.502374</td>\n    </tr>\n    <tr>\n      <th>26</th>\n      <td>q05_roll_std_4096</td>\n      <td>20740.235864</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"Trying the first 3 earthquakes only:"},{"metadata":{"trusted":true},"cell_type":"code","source":"params['num_boost_round'] = 99999\nparams['num_class'] = 3\nidx = y_tr[(y_tr.quake_number >= 0) & (y_tr.quake_number < 3)].index\ndataset = lgb.Dataset(X_tr.iloc[idx], label=y_tr.loc[idx, 'quake_number'])\nresult = lgb.cv(params, dataset, nfold=10, early_stopping_rounds=100, stratified=False)\nplot_multiclass(result)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<h2>Binary prediction</h2>\n\nWe can also try a binary classification model with only two earthquakes. In this case, the results are very different depending on the quakes we are comparing:"},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"params = {\n    'objective': 'binary',\n    'metric': 'auc',\n    \"boosting\": \"gbdt\",\n    'num_leaves': 22,\n    'min_data_in_leaf': 10, \n    'max_depth': -1,\n    'learning_rate': 0.01,\n    \"feature_fraction\": 1,\n    \"bagging_freq\": 5,\n    \"bagging_fraction\": 0.9,\n    \"bagging_seed\": 19,\n    \"lambda_l2\": 0.05,\n    \"num_boost_round\": 90000,\n    \"verbosity\": -1,\n    \"nthread\": -1,\n}\n# Predict if a segment came from group A or B\ndef binary_prediction(q1, q2):\n    assert(q1 < q2)\n    idx = y_tr[(y_tr.quake_number == q1) | (y_tr.quake_number == q2)].index\n    binary_target = y_tr.loc[idx, 'quake_number'] > q1\n    dataset = lgb.Dataset(X_tr.iloc[idx], label=binary_target)\n    result = lgb.cv(params, dataset, nfold=10, early_stopping_rounds=100, stratified=True)\n    num_rounds = len(result['auc-mean'])\n\n    plt.figure(figsize=(10, 5))\n    plt.title(\"AUC - earthquake {} vs {}\".format(q1, q2))\n    ax = sns.lineplot(x=np.arange(num_rounds), y=result['auc-mean'])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"binary_prediction(2, 3)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"binary_prediction(2, 11)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"binary_prediction(1, 6)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"<h2>Predict TTF > 12</h2>\n\nMost of the error is coming from long earthquake cycles, we can try a classifier to distinguish between long and short cycles."},{"metadata":{"trusted":true},"cell_type":"code","source":"binary_target = (y_tr.time_to_failure > 12).astype('int8')\nbinary_target.value_counts()","execution_count":null,"outputs":[]},{"metadata":{"_kg_hide-input":true,"trusted":true},"cell_type":"code","source":"f = X_tr.columns\ndataset = lgb.Dataset(X_tr[f], label=binary_target)\nresult = lgb.cv(params, dataset, nfold=10, early_stopping_rounds=100,\n                stratified=False, shuffle=True)\nnum_rounds = len(result['auc-mean'])\n\nplt.figure(figsize=(10, 5))\nplt.title(\"AUC - predicting TTF > 12\")\nax = sns.lineplot(x=np.arange(num_rounds), y=result['auc-mean'])","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"With only two features:"},{"metadata":{"_kg_hide-input":false,"trusted":true},"cell_type":"code","source":"f = ['q05_roll_std_64', 'q95_roll_mean_64']\ndataset = lgb.Dataset(X_tr[f], label=binary_target)\nresult = lgb.cv(params, dataset, nfold=10, early_stopping_rounds=100,\n                stratified=False, shuffle=True)\nnum_rounds = len(result['auc-mean'])\n\nplt.figure(figsize=(10, 5))\nplt.title(\"AUC - predicting TTF > 12\")\nax = sns.lineplot(x=np.arange(num_rounds), y=result['auc-mean'])","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.4","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}