http://git-wip-us.apache.org/repos/asf/madlib-site/blob/acd339f6/community-artifacts/Random-forest-v2.ipynb ---------------------------------------------------------------------- diff --git a/community-artifacts/Random-forest-v2.ipynb b/community-artifacts/Random-forest-v2.ipynb new file mode 100644 index 0000000..87605b7 --- /dev/null +++ b/community-artifacts/Random-forest-v2.ipynb @@ -0,0 +1,3082 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Random forest\n", + "\n", + "Random forests build an ensemble of classifiers, each of which is a tree model constructed using bootstrapped samples from the input data. The results of these models are then combined to yield a single prediction, which, at the expense of some loss in interpretation, have been found to be highly accurate.\n", + "\n", + "Please also refer to the decision tree user documentation for information relevant to the implementation of random forests in MADlib.\n", + "\n", + "This notebook includes impurity importance which was added in 1.15." + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/fmcquillan/anaconda/lib/python2.7/site-packages/IPython/config.py:13: ShimWarning: The `IPython.config` package has been deprecated. You should import from traitlets.config instead.\n", + " \"You should import from traitlets.config instead.\", ShimWarning)\n", + "/Users/fmcquillan/anaconda/lib/python2.7/site-packages/IPython/utils/traitlets.py:5: UserWarning: IPython.utils.traitlets has moved to a top-level traitlets package.\n", + " warn(\"IPython.utils.traitlets has moved to a top-level traitlets package.\")\n" + ] + } + ], + "source": [ + "%load_ext sql" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "u'Connected: gpadmin@madlib'" + ] + }, + "execution_count": 2, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "# Greenplum Database 5.4.0 on GCP (demo machine)\n", + "%sql postgresql://gpadmin@35.184.253.255:5432/madlib\n", + " \n", + "# PostgreSQL local\n", + "#%sql postgresql://fmcquillan@localhost:5432/madlib" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "1 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>version</th>\n", + " </tr>\n", + " <tr>\n", + " <td>MADlib version: 1.15-dev, git revision: rc/1.14-rc1-45-g3ab7554, cmake configuration time: Wed Aug 1 18:34:10 UTC 2018, build type: release, build system: Linux-2.6.32-696.20.1.el6.x86_64, C compiler: gcc 4.4.7, C++ compiler: g++ 4.4.7</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(u'MADlib version: 1.15-dev, git revision: rc/1.14-rc1-45-g3ab7554, cmake configuration time: Wed Aug 1 18:34:10 UTC 2018, build type: release, build system: Linux-2.6.32-696.20.1.el6.x86_64, C compiler: gcc 4.4.7, C++ compiler: g++ 4.4.7',)]" + ] + }, + "execution_count": 3, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%sql select madlib.version();\n", + "#%sql select version();" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Random forest classification examples" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# 1. Load data\n", + "Data set related to whether to play golf or not." + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Done.\n", + "Done.\n", + "14 rows affected.\n", + "14 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>id</th>\n", + " <th>OUTLOOK</th>\n", + " <th>temperature</th>\n", + " <th>humidity</th>\n", + " <th>Temp_Humidity</th>\n", + " <th>clouds_airquality</th>\n", + " <th>windy</th>\n", + " <th>class</th>\n", + " </tr>\n", + " <tr>\n", + " <td>1</td>\n", + " <td>sunny</td>\n", + " <td>85.0</td>\n", + " <td>85.0</td>\n", + " <td>[85.0, 85.0]</td>\n", + " <td>[u'none', u'unhealthy']</td>\n", + " <td>False</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>2</td>\n", + " <td>sunny</td>\n", + " <td>80.0</td>\n", + " <td>90.0</td>\n", + " <td>[80.0, 90.0]</td>\n", + " <td>[u'none', u'moderate']</td>\n", + " <td>True</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>3</td>\n", + " <td>overcast</td>\n", + " <td>83.0</td>\n", + " <td>78.0</td>\n", + " <td>[83.0, 78.0]</td>\n", + " <td>[u'low', u'moderate']</td>\n", + " <td>False</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>4</td>\n", + " <td>rain</td>\n", + " <td>70.0</td>\n", + " <td>96.0</td>\n", + " <td>[70.0, 96.0]</td>\n", + " <td>[u'low', u'moderate']</td>\n", + " <td>False</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>5</td>\n", + " <td>rain</td>\n", + " <td>68.0</td>\n", + " <td>80.0</td>\n", + " <td>[68.0, 80.0]</td>\n", + " <td>[u'medium', u'good']</td>\n", + " <td>False</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>6</td>\n", + " <td>rain</td>\n", + " <td>65.0</td>\n", + " <td>70.0</td>\n", + " <td>[65.0, 70.0]</td>\n", + " <td>[u'low', u'unhealthy']</td>\n", + " <td>True</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>7</td>\n", + " <td>overcast</td>\n", + " <td>64.0</td>\n", + " <td>65.0</td>\n", + " <td>[64.0, 65.0]</td>\n", + " <td>[u'medium', u'moderate']</td>\n", + " <td>True</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>8</td>\n", + " <td>sunny</td>\n", + " <td>72.0</td>\n", + " <td>95.0</td>\n", + " <td>[72.0, 95.0]</td>\n", + " <td>[u'high', u'unhealthy']</td>\n", + " <td>False</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>9</td>\n", + " <td>sunny</td>\n", + " <td>69.0</td>\n", + " <td>70.0</td>\n", + " <td>[69.0, 70.0]</td>\n", + " <td>[u'high', u'good']</td>\n", + " <td>False</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>10</td>\n", + " <td>rain</td>\n", + " <td>75.0</td>\n", + " <td>80.0</td>\n", + " <td>[75.0, 80.0]</td>\n", + " <td>[u'medium', u'good']</td>\n", + " <td>False</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>11</td>\n", + " <td>sunny</td>\n", + " <td>75.0</td>\n", + " <td>70.0</td>\n", + " <td>[75.0, 70.0]</td>\n", + " <td>[u'none', u'good']</td>\n", + " <td>True</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>12</td>\n", + " <td>overcast</td>\n", + " <td>72.0</td>\n", + " <td>90.0</td>\n", + " <td>[72.0, 90.0]</td>\n", + " <td>[u'medium', u'moderate']</td>\n", + " <td>True</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>13</td>\n", + " <td>overcast</td>\n", + " <td>81.0</td>\n", + " <td>75.0</td>\n", + " <td>[81.0, 75.0]</td>\n", + " <td>[u'medium', u'moderate']</td>\n", + " <td>False</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>14</td>\n", + " <td>rain</td>\n", + " <td>71.0</td>\n", + " <td>80.0</td>\n", + " <td>[71.0, 80.0]</td>\n", + " <td>[u'low', u'unhealthy']</td>\n", + " <td>True</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(1, u'sunny', 85.0, 85.0, [85.0, 85.0], [u'none', u'unhealthy'], False, u\"Don't Play\"),\n", + " (2, u'sunny', 80.0, 90.0, [80.0, 90.0], [u'none', u'moderate'], True, u\"Don't Play\"),\n", + " (3, u'overcast', 83.0, 78.0, [83.0, 78.0], [u'low', u'moderate'], False, u'Play'),\n", + " (4, u'rain', 70.0, 96.0, [70.0, 96.0], [u'low', u'moderate'], False, u'Play'),\n", + " (5, u'rain', 68.0, 80.0, [68.0, 80.0], [u'medium', u'good'], False, u'Play'),\n", + " (6, u'rain', 65.0, 70.0, [65.0, 70.0], [u'low', u'unhealthy'], True, u\"Don't Play\"),\n", + " (7, u'overcast', 64.0, 65.0, [64.0, 65.0], [u'medium', u'moderate'], True, u'Play'),\n", + " (8, u'sunny', 72.0, 95.0, [72.0, 95.0], [u'high', u'unhealthy'], False, u\"Don't Play\"),\n", + " (9, u'sunny', 69.0, 70.0, [69.0, 70.0], [u'high', u'good'], False, u'Play'),\n", + " (10, u'rain', 75.0, 80.0, [75.0, 80.0], [u'medium', u'good'], False, u'Play'),\n", + " (11, u'sunny', 75.0, 70.0, [75.0, 70.0], [u'none', u'good'], True, u'Play'),\n", + " (12, u'overcast', 72.0, 90.0, [72.0, 90.0], [u'medium', u'moderate'], True, u'Play'),\n", + " (13, u'overcast', 81.0, 75.0, [81.0, 75.0], [u'medium', u'moderate'], False, u'Play'),\n", + " (14, u'rain', 71.0, 80.0, [71.0, 80.0], [u'low', u'unhealthy'], True, u\"Don't Play\")]" + ] + }, + "execution_count": 5, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "DROP TABLE IF EXISTS rf_golf CASCADE;\n", + "\n", + "CREATE TABLE rf_golf (\n", + " id integer NOT NULL,\n", + " \"OUTLOOK\" text,\n", + " temperature double precision,\n", + " humidity double precision,\n", + " \"Temp_Humidity\" double precision[],\n", + " clouds_airquality text[],\n", + " windy boolean,\n", + " class text\n", + ");\n", + "\n", + "INSERT INTO rf_golf VALUES\n", + "(1,'sunny', 85, 85, ARRAY[85, 85],ARRAY['none', 'unhealthy'], 'false','Don''t Play'),\n", + "(2, 'sunny', 80, 90, ARRAY[80, 90], ARRAY['none', 'moderate'], 'true', 'Don''t Play'),\n", + "(3, 'overcast', 83, 78, ARRAY[83, 78], ARRAY['low', 'moderate'], 'false', 'Play'),\n", + "(4, 'rain', 70, 96, ARRAY[70, 96], ARRAY['low', 'moderate'], 'false', 'Play'),\n", + "(5, 'rain', 68, 80, ARRAY[68, 80], ARRAY['medium', 'good'], 'false', 'Play'),\n", + "(6, 'rain', 65, 70, ARRAY[65, 70], ARRAY['low', 'unhealthy'], 'true', 'Don''t Play'),\n", + "(7, 'overcast', 64, 65, ARRAY[64, 65], ARRAY['medium', 'moderate'], 'true', 'Play'),\n", + "(8, 'sunny', 72, 95, ARRAY[72, 95], ARRAY['high', 'unhealthy'], 'false', 'Don''t Play'),\n", + "(9, 'sunny', 69, 70, ARRAY[69, 70], ARRAY['high', 'good'], 'false', 'Play'),\n", + "(10, 'rain', 75, 80, ARRAY[75, 80], ARRAY['medium', 'good'], 'false', 'Play'),\n", + "(11, 'sunny', 75, 70, ARRAY[75, 70], ARRAY['none', 'good'], 'true', 'Play'),\n", + "(12, 'overcast', 72, 90, ARRAY[72, 90], ARRAY['medium', 'moderate'], 'true', 'Play'),\n", + "(13, 'overcast', 81, 75, ARRAY[81, 75], ARRAY['medium', 'moderate'], 'false', 'Play'),\n", + "(14, 'rain', 71, 80, ARRAY[71, 80], ARRAY['low', 'unhealthy'], 'true', 'Don''t Play');\n", + "\n", + "SELECT * FROM rf_golf ORDER BY id;" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# 2. Train random forest\n", + "Train random forest and view the summary table:" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Done.\n", + "1 rows affected.\n", + "1 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>method</th>\n", + " <th>is_classification</th>\n", + " <th>source_table</th>\n", + " <th>model_table</th>\n", + " <th>id_col_name</th>\n", + " <th>dependent_varname</th>\n", + " <th>independent_varnames</th>\n", + " <th>cat_features</th>\n", + " <th>con_features</th>\n", + " <th>grouping_cols</th>\n", + " <th>num_trees</th>\n", + " <th>num_random_features</th>\n", + " <th>max_tree_depth</th>\n", + " <th>min_split</th>\n", + " <th>min_bucket</th>\n", + " <th>num_splits</th>\n", + " <th>verbose</th>\n", + " <th>importance</th>\n", + " <th>num_permutations</th>\n", + " <th>num_all_groups</th>\n", + " <th>num_failed_groups</th>\n", + " <th>total_rows_processed</th>\n", + " <th>total_rows_skipped</th>\n", + " <th>dependent_var_levels</th>\n", + " <th>dependent_var_type</th>\n", + " <th>independent_var_types</th>\n", + " <th>null_proxy</th>\n", + " </tr>\n", + " <tr>\n", + " <td>forest_train</td>\n", + " <td>True</td>\n", + " <td>rf_golf</td>\n", + " <td>train_output</td>\n", + " <td>id</td>\n", + " <td>class</td>\n", + " <td>\"OUTLOOK\",windy,temperature,humidity</td>\n", + " <td>\"OUTLOOK\",windy</td>\n", + " <td>temperature,humidity</td>\n", + " <td>None</td>\n", + " <td>20</td>\n", + " <td>2</td>\n", + " <td>8</td>\n", + " <td>3</td>\n", + " <td>1</td>\n", + " <td>10</td>\n", + " <td>False</td>\n", + " <td>True</td>\n", + " <td>1</td>\n", + " <td>1</td>\n", + " <td>0</td>\n", + " <td>14</td>\n", + " <td>0</td>\n", + " <td>\"Don't Play\",\"Play\"</td>\n", + " <td>text</td>\n", + " <td>text, boolean, double precision, double precision</td>\n", + " <td>None</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(u'forest_train', True, u'rf_golf', u'train_output', u'id', u'class', u'\"OUTLOOK\",windy,temperature,humidity', u'\"OUTLOOK\",windy', u'temperature,humidity', None, 20, 2, 8, 3, 1, 10, False, True, 1, 1, 0, 14, 0, u'\"Don\\'t Play\",\"Play\"', u'text', u'text, boolean, double precision, double precision', u'None')]" + ] + }, + "execution_count": 6, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "DROP TABLE IF EXISTS train_output, train_output_group, train_output_summary;\n", + "\n", + "SELECT madlib.forest_train('rf_golf', -- source table\n", + " 'train_output', -- output model table\n", + " 'id', -- id column\n", + " 'class', -- response\n", + " '\"OUTLOOK\", temperature, humidity, windy', -- features\n", + " NULL, -- exclude columns\n", + " NULL, -- grouping columns\n", + " 20::integer, -- number of trees\n", + " 2::integer, -- number of random features\n", + " TRUE::boolean, -- variable importance\n", + " 1::integer, -- num_permutations\n", + " 8::integer, -- max depth\n", + " 3::integer, -- min split\n", + " 1::integer, -- min bucket\n", + " 10::integer -- number of splits per continuous variable\n", + " );\n", + "\n", + "SELECT * FROM train_output_summary;" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Review the group table for variable importance:" + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "1 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>gid</th>\n", + " <th>success</th>\n", + " <th>cat_n_levels</th>\n", + " <th>cat_levels_in_text</th>\n", + " <th>oob_error</th>\n", + " <th>oob_var_importance</th>\n", + " <th>impurity_var_importance</th>\n", + " </tr>\n", + " <tr>\n", + " <td>1</td>\n", + " <td>True</td>\n", + " <td>[3, 2]</td>\n", + " <td>[u'overcast', u'sunny', u'rain', u'False', u'True']</td>\n", + " <td>0.28571428571428571429</td>\n", + " <td>[0.06625, 0.0233928571428572, 0.0, 0.0145833333333333]</td>\n", + " <td>[0.27116534244345, 0.10438655409181, 0.307256009757555, 0.184889160052295]</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(1, True, [3, 2], [u'overcast', u'sunny', u'rain', u'False', u'True'], Decimal('0.28571428571428571429'), [0.06625, 0.0233928571428572, 0.0, 0.0145833333333333], [0.27116534244345, 0.10438655409181, 0.307256009757555, 0.184889160052295])]" + ] + }, + "execution_count": 7, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "SELECT * FROM train_output_group;" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "The 'cat_levels_in_text' array shows the levels of the categorical variables \"OUTLOOK\" and windy, which have 3 and 2 levels respectively. \n", + "\n", + "Out-of-bag and impurity variable importance arrays are ordered according to the order of the variables in 'independent_varnames' in model summary table. A higher value means higher importance for the variable. \n", + "\n", + "We can use the helper function to get a normalized view of variable importance:" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Done.\n", + "1 rows affected.\n", + "4 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>feature</th>\n", + " <th>oob_var_importance</th>\n", + " <th>impurity_var_importance</th>\n", + " </tr>\n", + " <tr>\n", + " <td>\"OUTLOOK\"</td>\n", + " <td>63.5636778983</td>\n", + " <td>31.2511535374</td>\n", + " </tr>\n", + " <tr>\n", + " <td>windy</td>\n", + " <td>22.4443175328</td>\n", + " <td>12.0302992992</td>\n", + " </tr>\n", + " <tr>\n", + " <td>humidity</td>\n", + " <td>13.9920045688</td>\n", + " <td>21.3080310195</td>\n", + " </tr>\n", + " <tr>\n", + " <td>temperature</td>\n", + " <td>0.0</td>\n", + " <td>35.4105161438</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(u'\"OUTLOOK\"', 63.5636778983438, 31.2511535374489),\n", + " (u'windy', 22.4443175328384, 12.0302992992133),\n", + " (u'humidity', 13.9920045688178, 21.3080310195216),\n", + " (u'temperature', 0.0, 35.4105161438162)]" + ] + }, + "execution_count": 9, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "DROP TABLE IF EXISTS imp_output;\n", + "\n", + "SELECT madlib.get_var_importance('train_output','imp_output');\n", + "SELECT * FROM imp_output ORDER BY oob_var_importance DESC;" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "scrolled": true + }, + "source": [ + "# 3. Predict\n", + "Predict output categories. For the purpose of this example, we use the same data that was used for training:" + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Done.\n", + "1 rows affected.\n", + "14 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>id</th>\n", + " <th>class</th>\n", + " <th>estimated_class</th>\n", + " </tr>\n", + " <tr>\n", + " <td>1</td>\n", + " <td>Don't Play</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>2</td>\n", + " <td>Don't Play</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>3</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>4</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>5</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>6</td>\n", + " <td>Don't Play</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>7</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>8</td>\n", + " <td>Don't Play</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>9</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>10</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>11</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>12</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>13</td>\n", + " <td>Play</td>\n", + " <td>Play</td>\n", + " </tr>\n", + " <tr>\n", + " <td>14</td>\n", + " <td>Don't Play</td>\n", + " <td>Don't Play</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(1, u\"Don't Play\", u\"Don't Play\"),\n", + " (2, u\"Don't Play\", u\"Don't Play\"),\n", + " (3, u'Play', u'Play'),\n", + " (4, u'Play', u'Play'),\n", + " (5, u'Play', u'Play'),\n", + " (6, u\"Don't Play\", u\"Don't Play\"),\n", + " (7, u'Play', u'Play'),\n", + " (8, u\"Don't Play\", u\"Don't Play\"),\n", + " (9, u'Play', u'Play'),\n", + " (10, u'Play', u'Play'),\n", + " (11, u'Play', u'Play'),\n", + " (12, u'Play', u'Play'),\n", + " (13, u'Play', u'Play'),\n", + " (14, u\"Don't Play\", u\"Don't Play\")]" + ] + }, + "execution_count": 10, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "DROP TABLE IF EXISTS prediction_results;\n", + "\n", + "SELECT madlib.forest_predict('train_output', -- tree model \n", + " 'rf_golf', -- new data table\n", + " 'prediction_results', -- output table\n", + " 'response'); -- show response\n", + "\n", + "SELECT g.id, class, estimated_class FROM prediction_results p, \n", + "rf_golf g WHERE p.id = g.id ORDER BY g.id;" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "To display the probabilities associated with each value of the dependent variable, set the 'type' parameter to 'prob':" + ] + }, + { + "cell_type": "code", + "execution_count": 11, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Done.\n", + "1 rows affected.\n", + "14 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>id</th>\n", + " <th>class</th>\n", + " <th>estimated_prob_Don't Play</th>\n", + " <th>estimated_prob_Play</th>\n", + " </tr>\n", + " <tr>\n", + " <td>1</td>\n", + " <td>Don't Play</td>\n", + " <td>0.95</td>\n", + " <td>0.05</td>\n", + " </tr>\n", + " <tr>\n", + " <td>2</td>\n", + " <td>Don't Play</td>\n", + " <td>0.9</td>\n", + " <td>0.1</td>\n", + " </tr>\n", + " <tr>\n", + " <td>3</td>\n", + " <td>Play</td>\n", + " <td>0.05</td>\n", + " <td>0.95</td>\n", + " </tr>\n", + " <tr>\n", + " <td>4</td>\n", + " <td>Play</td>\n", + " <td>0.25</td>\n", + " <td>0.75</td>\n", + " </tr>\n", + " <tr>\n", + " <td>5</td>\n", + " <td>Play</td>\n", + " <td>0.2</td>\n", + " <td>0.8</td>\n", + " </tr>\n", + " <tr>\n", + " <td>6</td>\n", + " <td>Don't Play</td>\n", + " <td>0.7</td>\n", + " <td>0.3</td>\n", + " </tr>\n", + " <tr>\n", + " <td>7</td>\n", + " <td>Play</td>\n", + " <td>0.1</td>\n", + " <td>0.9</td>\n", + " </tr>\n", + " <tr>\n", + " <td>8</td>\n", + " <td>Don't Play</td>\n", + " <td>0.8</td>\n", + " <td>0.2</td>\n", + " </tr>\n", + " <tr>\n", + " <td>9</td>\n", + " <td>Play</td>\n", + " <td>0.05</td>\n", + " <td>0.95</td>\n", + " </tr>\n", + " <tr>\n", + " <td>10</td>\n", + " <td>Play</td>\n", + " <td>0.3</td>\n", + " <td>0.7</td>\n", + " </tr>\n", + " <tr>\n", + " <td>11</td>\n", + " <td>Play</td>\n", + " <td>0.15</td>\n", + " <td>0.85</td>\n", + " </tr>\n", + " <tr>\n", + " <td>12</td>\n", + " <td>Play</td>\n", + " <td>0.4</td>\n", + " <td>0.6</td>\n", + " </tr>\n", + " <tr>\n", + " <td>13</td>\n", + " <td>Play</td>\n", + " <td>0.05</td>\n", + " <td>0.95</td>\n", + " </tr>\n", + " <tr>\n", + " <td>14</td>\n", + " <td>Don't Play</td>\n", + " <td>0.85</td>\n", + " <td>0.15</td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(1, u\"Don't Play\", 0.95, 0.05),\n", + " (2, u\"Don't Play\", 0.9, 0.1),\n", + " (3, u'Play', 0.05, 0.95),\n", + " (4, u'Play', 0.25, 0.75),\n", + " (5, u'Play', 0.2, 0.8),\n", + " (6, u\"Don't Play\", 0.7, 0.3),\n", + " (7, u'Play', 0.1, 0.9),\n", + " (8, u\"Don't Play\", 0.8, 0.2),\n", + " (9, u'Play', 0.05, 0.95),\n", + " (10, u'Play', 0.3, 0.7),\n", + " (11, u'Play', 0.15, 0.85),\n", + " (12, u'Play', 0.4, 0.6),\n", + " (13, u'Play', 0.05, 0.95),\n", + " (14, u\"Don't Play\", 0.85, 0.15)]" + ] + }, + "execution_count": 11, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "DROP TABLE IF EXISTS prediction_results;\n", + "\n", + "SELECT madlib.forest_predict('train_output', -- tree model \n", + " 'rf_golf', -- new data table\n", + " 'prediction_results', -- output table\n", + " 'prob'); -- show probability\n", + "\n", + "SELECT g.id, class, \"estimated_prob_Don't Play\", \"estimated_prob_Play\" \n", + "FROM prediction_results p, rf_golf g WHERE p.id = g.id ORDER BY g.id;" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# 4. View tree in text format\n", + "View a single tree within the forest identified by 'gid' and 'sample_id', out of the several that were created:" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "1 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>get_tree</th>\n", + " </tr>\n", + " <tr>\n", + " <td>-------------------------------------<br> - Each node represented by 'id' inside ().<br> - Each internal nodes has the split condition at the end, while each<br> leaf node has a * at the end.<br> - For each internal node (i), its child nodes are indented by 1 level<br> with ids (2i+1) for True node and (2i+2) for False node.<br> - Number of (weighted) rows for each response variable inside [].'<br> The response label order is given as ['\"Don\\'t Play\"', '\"Play\"'].<br> For each leaf, the prediction is given after the '-->'<br> <br>-------------------------------------<br>(0)[ 4 12] temperature <= 65<br> (1)[2 0] * --> \"Don't Play\"<br> (2)[ 2 12] humidity <= 78<br> (5)[0 8] * --> \"Play\"<br> (6)[2 4] windy in {False}<br> (13)[1 4] temperature <= 70<br> (27)[0 3] * --> \"Play\"<br> (28)[1 1] * --> \"Don't Play\"<br> (14)[1 0] * - -> \"Don't Play\"<br></td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(u'-------------------------------------\\n - Each node represented by \\'id\\' inside ().\\n - Each internal nodes has the split condition at the end, while each\\n leaf node has a * at the end.\\n - For each internal node (i), its child nodes are indented by 1 level\\n with ids (2i+1) for True node and (2i+2) for False node.\\n - Number of (weighted) rows for each response variable inside [].\\'\\n The response label order is given as [\\'\"Don\\\\\\'t Play\"\\', \\'\"Play\"\\'].\\n For each leaf, the prediction is given after the \\'-->\\'\\n \\n-------------------------------------\\n(0)[ 4 12] temperature <= 65\\n (1)[2 0] * --> \"Don\\'t Play\"\\n (2)[ 2 12] humidity <= 78\\n (5)[0 8] * --> \"Play\"\\n (6)[2 4] windy in {False}\\n (13)[1 4] temperature <= 70\\n (27)[0 3] * --> \"Play\"\\n (28)[1 1] * --> \"Don\\'t Play\"\\n (14)[1 0] * --> \"Don\\'t Play\"\\n',) ]" + ] + }, + "execution_count": 9, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "SELECT madlib.get_tree('train_output',1,7, FALSE);" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Please see the decision tree user docs for an explanation on how to interpret the tree display above." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# 5. View tree in dot format" + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "1 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>get_tree</th>\n", + " </tr>\n", + " <tr>\n", + " <td>digraph \"Classification tree for rf_golf\" {<br>\"0\" [label=\"temperature <= 65\", shape=ellipse];<br>\"0\" -> \"1\"[label=\"yes\"];<br>\"1\" [label=\"\\\"Don't Play\\\"\",shape=box];<br>\"0\" -> \"2\"[label=\"no\"];<br>\"2\" [label=\"humidity <= 78\", shape=ellipse];<br>\"2\" -> \"5\"[label=\"yes\"];<br>\"5\" [label=\"\\\"Play\\\"\",shape=box];<br>\"2\" -> \"6\"[label=\"no\"];<br>\"6\" [label=\"windy <= False\", shape=ellipse];<br>\"6\" -> \"13\"[label=\"yes\"];<br>\"6\" -> \"14\"[label=\"no\"];<br>\"14\" [label=\"\\\"Don't Play\\\"\",shape=box];<br>\"13\" [label=\"temperature <= 70\", shape=ellipse];<br>\"13\" -> \"27\"[label=\"yes\"];<br>\"27\" [label=\"\\\"Play\\\"\",shape=box];<br>\"13\" -> \"28\"[label=\"no\"];<br>\"28\" [label=\"\\\"Don't Play\\\"\",shape=box];<br><br>} //---end of digraph--------- </td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(u'digraph \"Classification tree for rf_golf\" {\\n\"0\" [label=\"temperature <= 65\", shape=ellipse];\\n\"0\" -> \"1\"[label=\"yes\"];\\n\"1\" [label=\"\\\\\"Don\\'t Play\\\\\"\",shape=box];\\n\"0\" -> \"2\"[label=\"no\"];\\n\"2\" [label=\"humidity <= 78\", shape=ellipse];\\n\"2\" -> \"5\"[label=\"yes\"];\\n\"5\" [label=\"\\\\\"Play\\\\\"\",shape=box];\\n\"2\" -> \"6\"[label=\"no\"];\\n\"6\" [label=\"windy <= False\", shape=ellipse];\\n\"6\" -> \"13\"[label=\"yes\"];\\n\"6\" -> \"14\"[label=\"no\"];\\n\"14\" [label=\"\\\\\"Don\\'t Play\\\\\"\",shape=box];\\n\"13\" [label=\"temperature <= 70\", shape=ellipse];\\n\"13\" -> \"27\"[label=\"yes\"];\\n\"27\" [label=\"\\\\\"Play\\\\\"\",shape=box];\\n\"13\" -> \"28\"[label=\"no\"];\\n\"28\" [label=\"\\\\\"Don\\'t Play\\\\\"\",shape=box];\\n\\n} //---end of digraph--------- ',)]" + ] + }, + "execution_count": 10, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "SELECT madlib.get_tree('train_output',1,7);" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# 6. View tree in dot format with additional information" + ] + }, + { + "cell_type": "code", + "execution_count": 11, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "1 rows affected.\n" + ] + }, + { + "data": { + "text/html": [ + "<table>\n", + " <tr>\n", + " <th>get_tree</th>\n", + " </tr>\n", + " <tr>\n", + " <td>digraph \"Classification tree for rf_golf\" {<br>\"0\" [label=\"temperature <= 65\\n impurity = 0.375\\n samples = 16\\n value = [ 4 12]\\n class = \\\"Play\\\"\", shape=ellipse];<br>\"0\" -> \"1\"[label=\"yes\"];<br>\"1\" [label=\"\\\"Don't Play\\\"\\n impurity = 0\\n samples = 2\\n value = [2 0]\",shape=box];<br>\"0\" -> \"2\"[label=\"no\"];<br>\"2\" [label=\"humidity <= 78\\n impurity = 0.244898\\n samples = 14\\n value = [ 2 12]\\n class = \\\"Play\\\"\", shape=ellipse];<br>\"2\" -> \"5\"[label=\"yes\"];<br>\"5\" [label=\"\\\"Play\\\"\\n impurity = 0\\n samples = 8\\n value = [0 8]\",shape=box];<br>\"2\" -> \"6\"[label=\"no\"];<br>\"6\" [label=\"windy <= False\\n impurity = 0.444444\\n samples = 6\\n value = [2 4]\\n class = \\\"Play\\\"\", shape=ellipse];<br>\"6\" -> \"13\"[label=\"yes\"];<br>\"6\" -> \"14\"[label=\"no\"];<br>\"14\" [label=\"\\\"Don't Play\\\"\\n impurity = 0\\n samples = 1\\n value = [1 0]\",shape=box];<br>\"13\" [label=\"temperature <= 70\\n impurity = 0.32\\n samples = 5\\n value = [1 4]\\n class = \\\"Play\\\"\", shape=ellipse];<br>\"13\" -> \"27\"[label=\"yes\"];<br>\"27\" [label=\"\\\"Play\\\"\\n impurity = 0\\n samples = 3\\n value = [0 3]\",shape=box];<br>\"13\" -> \"28\"[label=\"no\"];<br>\"28\" [label=\"\\\"Don't Play\\\"\\n impurity = 0.5\\n samples = 2\\n value = [1 1]\",shape=box];<br><br>} //---end of digraph--------- </td>\n", + " </tr>\n", + "</table>" + ], + "text/plain": [ + "[(u'digraph \"Classification tree for rf_golf\" {\\n\"0\" [label=\"temperature <= 65\\\\n impurity = 0.375\\\\n samples = 16\\\\n value = [ 4 12]\\\\n class = \\\\\"Play\\\\\"\", shape=ellipse];\\n\"0\" -> \"1\"[label=\"yes\"];\\n\"1\" [label=\"\\\\\"Don\\'t Play\\\\\"\\\\n impurity = 0\\\\n samples = 2\\\\n value = [2 0]\",shape=box];\\n\"0\" -> \"2\"[label=\"no\"];\\n\"2\" [label=\"humidity <= 78\\\\n impurity = 0.244898\\\\n samples = 14\\\\n value = [ 2 12]\\\\n class = \\\\\"Play\\\\\"\", shape=ellipse];\\n\"2\" -> \"5\"[label=\"yes\"];\\n\"5\" [label=\"\\\\\"Play\\\\\"\\\\n impurity = 0\\\\n samples = 8\\\\n value = [0 8]\",shape=box];\\n\"2\" -> \"6\"[label=\"no\"];\\n\"6\" [label=\"windy <= False\\\\n impurity = 0.444444\\\\n samples = 6\\\\n value = [2 4]\\\\n class = \\\\\"Play\\\\\"\", shape=ellipse];\\n\"6\" -> \"13\"[label=\"yes\"];\\n\"6\" -> \"14\"[label=\"no\"];\\n\"14\" [label=\"\\\\\"Don\\'t Play\\\\\"\\\\n impurity = 0\\\\n samples = 1\\\\n value = [1 0]\" ,shape=box];\\n\"13\" [label=\"temperature <= 70\\\\n impurity = 0.32\\\\n samples = 5\\\\n value = [1 4]\\\\n class = \\\\\"Play\\\\\"\", shape=ellipse];\\n\"13\" -> \"27\"[label=\"yes\"];\\n\"27\" [label=\"\\\\\"Play\\\\\"\\\\n impurity = 0\\\\n samples = 3\\\\n value = [0 3]\",shape=box];\\n\"13\" -> \"28\"[label=\"no\"];\\n\"28\" [label=\"\\\\\"Don\\'t Play\\\\\"\\\\n impurity = 0.5\\\\n samples = 2\\\\n value = [1 1]\",shape=box];\\n\\n} //---end of digraph--------- ',)]" + ] + }, + "execution_count": 11, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "%%sql\n", + "SELECT madlib.get_tree('train_output',1,7, TRUE, TRUE);" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "You may wish to visualize the tree using pygraphviz or another program that can handle dot format:" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "1 rows affected.\n", + "True\n" + ] + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkQAAAO3CAYAAADCkVtZAAAAAXNSR0IArs4c6QAAQABJREFUeAHs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\nXxDB119/3ckg72iZ/vKXvzhJ6tmzZ+C11lpruSapN dqtVtwyXSoEhIAQEAJCIC8CIkR5Ycl/EL+e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\nEJEz6I477ggXXnih1w3DN+jY Y4/1aLHGQND56kEAk+hll10WbrrpJi98O2DAANcaLbjggtWzSK1E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\ns8967TF VKW/8pqlF+RDYdNNNAw7+nTt3DmiOKAciEQJCQAgIgepDoGKECBPZhhtuGHr27BleffXV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 9+233w5v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\nC9OL5W3yviEoEIIXX3wxM1buejkBickmRByj75gQsV9ozIEDB0b mrEwTl8mTJ/uYppmJDxV8//e/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\nDq/BnEgDOXEwM1VKePgQ2RPLKqus4hod8+ eJD/k72pCrrrqqzrFi7Zhvkpu4IDGtGZ/AP4hcnIE5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\njljDhIbZpqnCmP/5z 3/chIaTeGsFfLLnXIz+MOGRADH2fYr7tLQFXmQV819z1hxfX+p3TLr4WoEJ\n/kTN9W0r9fzUvxAQAkKglhFoESF67LHH3IEWQmS/zGsSv2xCVJMAaNEtQoCIt27dugXTYHnKghZ1\noouEgBAQAkKg6Ag026kajYOZi4IlBqxZMsRdMJNhoBSI+eoU/aaow+pFgMAD82kL5g/m5VGqd6Va\nmRAQAkIgXQg0mxDxRU5oufnppGulRZytFfEMjzzyCP5XwQp5BhIUSoRAUxE4+OCDXUuUnQSzqdeq\nnRAQAkJACJQGgWaZzMiPQ7FO8s6YI3FpZpSCXvHdyXa9IpFjpZJRpgAuTTEPAuRmIls6ZmcrA5On\nhQ4JASEgBIRAORFoFiGyqCHPDPz666+nziEUR2HKTEDmyF2TBqGcBs7BFmKf2OlCkknQCCGkdItF\nzDVprlb4NljhVc8mbgVZg6U6yFyHY/c999yT2c/ewMncwvqzDwVLTxAsajBzDCd3izQLxXAKz3Ra\ngg380O69914vJZP0uZZg+epSCAgBIZAsBEzT0SR5+eWXPazZHlRNap+0RsOHD/f5V6L6eUuxoHo9\nRWaTKqQjsIzaHs5PJXtC/a2OW6PT5TpSIBxwwAFefNWixaL7778/c50l+fR7Zf8p9d5JbZAtFBS2\nCMI67ah4nwaheK8RIc+TlYb5ao5CQAgIgWpGoMl5iEiKt+aaa6YaC5IcpklIrki+p1yJcyHlHi/n\nvhVyjSAyJHCMhVxHbdu2jUxDEx+q907iTHI+xULuJKtTFm200UbxIc+dZBokzwtl5VCi+GVZsaOb\nb745046N/fff3xNJWmqDiBc5m0zDVKdNknfI9QRmcbLQJM9VcxMCQkAIVDMCTTKZWaJAz5+CIzH1\nqiSVQ8ASMnp0H0VDKynUWiPCzjSHmWkQgWjJK4MleAynn3565nj2BvXlunTpkn0o7L333m7yosQF\n OXowyXbv3r1OG0pfWLLK8Nlnn4U2bdr4OZz7iXakLEhazKB1FmU7lnnccz4dc8wx4dRTT809rX0h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 K+tx9imbSrULLHH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\nkEcHjVpMiHjAMx7aDa7nfdq0aYEyIuSfQWKzU/a84v6Y F348jBMLIftooejDosoy13P+wAMPDOYk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\nansRci9JNwIQoldffTXdi9DshYA QEAIpQyCvDxGmMqKZzIE4Zcspz3Tx4YEMxYLpppLkEU0VZqg4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\nKs763NCacu 87ZO2PP/7IPVxnf6mllqqXjbpOg0Z27rjjDo+UY619+vQJL774oicnBF9y8tSiiBDV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 UUZl5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\n0tFqRIA8UmgDRYiq8e5qTUJACCQVgbyEaNVVV/VQ86ROutrnRa4j8h thzTzhhBO8jlq1r1nr+x8C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\nHOmankIJH1syz+OPP95xzL4WvzUK1NZatmr+1 5ZbbrnQvXv3bDi0LQSEgBAQAmVAoKCGiLF32223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\niGnhP4JjMA+rONqqMtMt 7aiYCA8++OBA5BMO0/fcc0+dAXlgEx0WZ+9uqmYI8xsmtkKCr8zSSy/t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\nVXJ 1hKVT9JRot2+//TaRc2wOPviigTV+QxSslQgBISAEhEDlEWiWySye7nrrrefRTc8//7zXPIuP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 Fb748S1JuxBB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\nwcxD4W9/+1t4+OGHvU5WWgCHwODkyxqa4yBuZpFgJkNPWIk m6b333quzZIgCdeDo87DDDgv77LNP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\nBIRAXQTKQogYkocHYeY4BFt+m7DNNt vUnYn2ag4BTEtEmuWWDkkDEPhxUdeN1AeY/NJC8tOAreYo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\nRnfddZcXUK3NW 6dVtwQByDWZpjfZZJPQsWNHjyIjc3qbNm1a0p2uEQJCQAgIgSpBIBWEKMYaM9rA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 bk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\nBanjneOVJiE4V0OITjjhhHD22WfHkOpdCAgBISAEhECjCIgQNQpR6RtgX oo1MNTl+uyzzzKkJCYp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\ntdcOt9xyS7j88svDFVdcUeTe1Z0QEAJCQAikCQGF 3afgbk2aNMnNOvIhKv7NIuKMcPyjjjrKfbS2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\nhvfffz/stddeYfHFFw+9evU qTsfqRQgIASEgBCqGwEyRScVG18ANIkD5iPXXXz9MmTIlTJgwIUye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\nRp8+fc Kxxx7rVdxNS+ERU3vvvXduM+03A4GVVlopmNnRM1njk3XGGWc0ejUZsJGY2OIjhpAEMlvW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 gBIRAkRAQ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\nYO+9994bxo4dG4444oiqW58WJASEgBBIMgIiRAm7O2iI5D+UsJ tSxun06NEj3HbbbR6Kf+mll5Zx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\nVMUkCzL2888/Zw/vZIg6cu3bt8+8OnTo4 Nu8L7XUUm7GhkxJhIAQqG4ERIgScn+JFvrxxx/lQ5SQ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\nZpttNieyG220Udh8 881dU5f0NAqVQ1AjC4HiIyBCVHxMW9QjTqoHH3xwOOGEE1p0vS6qXQTQAqF9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\ nPCs5mckxyZJBWyIEhMD/EBAh+h8WZdladdVVvWwCkUL9+vUL+A9Nnz49/PDDD+GTTz7Rl1RZ7kJl\nBiEzMeHSq6++umcmJnIMMnzMMceoplVlbklNjEpBWSISKeCL8/VCCy0U0BaRrBPH/bjkS02AoUUK\ngQYQkFN1A+CU4hSRQrfddpuXWKCW0SyzzBLIL0PJBYT9Dh06eGVsIkhI5y9JNwLUFsO/g/QK66yz\njm+TZE8iBCqFwLvvvuuZzf/1r3+FZZdd1jVJffv29fpqlZqTxhUClUZAGqIy3wGqXUOEEFTWVB+P\nyRDHqJaNbwl+JSJDIJJeeeWVVzw3DAVUSZQ4ZswY/4UuMpTee1otM+/SpYundKCGGt9J/FBbccUV\n3feoWtaodQiB5iIgQtRcxFrZnszCjamo0RLhZCtJJwJUoSdfEPf622+/zRAhaopJhECSECC/FaZb\niBGm3F133dULzD799NNJmqbmIgTKgoAIUVlg/t8gOM1CeAoJ5w488EA3mxVqo+PJRIDweYrz8uv7\nhRde8F/bmMtEhJJ5vzSr/yEAMcKfCK0mhYHXXXdd9zOaOHHi/xppSwhUOQIiRGW+wZjBUE0XEsJm\nTzrppEKndTyhCJAQr3Pnzp4/6IILLggTJkxwR9aETlfTEgJ5ESDoA9PuAw88ECBDfFedeOKJ4ccf\nf8zbXgeFQDUhIEJUgbuJYy3Vq3MF7dBBBx0U2rdvn3tK+wlFgLIa6623Xth7772dAPEQIfdLvvub\n0CVoWkKgHgKbbbaZ58S6+OKLw9VXX+1JQ0eNGlWvnQ4IgWpCQISoAneTkPt8fkQ4W/NrTJJ8BHCS\nJgMwv6jJLfXiiy+GK6+8MiywwALJn7xmKASagADh+ocffnggIg3Sv+2224Ytt9wyTJ48uQlXq4kQ\nSB8CIkQVuGc42+YmRUM7RO0hkqlJko0AmaS7desWKMo7aNCggJ8QDqkSIVCN CCy66KKeKmTcuHFe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\nTEmahCK7aDWoTI9PBOaANP1S79ChQzj22GOdEFUa9x1 33DF8+eWXYaGFFio4FdoMHz48vPzyy3Xa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\nW/zYnnvumemfDSN2Uf/+/SMjZH WOF9oxwhcZEfE1GaGLjPRlmjLPp556qsGXORNn2je2YU7u0eDB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\nwRzykyvmC xXMnOQZ6zmHlog0BBAjMqUvsMACuZekdh9SBBk1k6iTIhJyEo0mEQJJQSB1hIgvCb6o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 OmzfAf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\nN3CvvfYKlC6ZZAEPEiFQMQTKraJqrVN1Y/NtyGQWX4s5xwCPLATYT VH4W+SqbPFvQJUbq7sxJ3AN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\n2oRIQRz3rRp7aqbOgzcOQEjipAkegGRaLqOw 7LLLBvNT8fIdjc0VDSpRT9wPNBcUZ4WsFBJKbuAY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\n+2ztE79oC2Xb5Us1V3N UirlVghBRqy63lEsp1lbKPrmXkIlcUkwWY45znlIY5mOS0SZwzMwhXi8L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\nP2 kdUlYKsHpBmUIZ9aSoaQoRRKtR91yaQkS/WP6YK4pt1OrGPu+yY15RISZQsmWjm/T6CGea9sV1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 F9eqrr3ZS51C3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\nglDGQoYd3+99991XvyswUIuVh12hgCtzAEsEnFhP+mOeSS IKjLaPZwjG23Id4ll5tIEyAmoCeJNI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\n2hDkCQEhFqU0IYiVTJdqf/QblywJ7 OJ9t2odPAi0FRbeVp2i8P1i+bzjjjvUuifxM8pt06lJYYEs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\nVjPbn4IAChH3 Wu7jJoZA1giMQzJd1p1W6+/ee+9V/h7eqmrxglTrJ+t91B6iNhr8QShLxF+QsZYk\nKGFkOQmpon4S79ObQFHKHnA8/cWFwpZC4NhW7g0sWMyfwPNmZcCAAY4gcGFZbraLqseRuecLhVZt\naDsLjQC8X7iuk+Lb8jQx7hu4ninYmhbrB9+YMEdrQWo4yXiwmzSGAJxoQrehL7k+m7exHqy1IZCO\ngMUQJWDjs64SdukmSOs8cV1vlAbff7XgV2INyDRrp+RJUU2bN27JuDUtra1tT0eA9yHIPp955hmH\nYo+iKSzS4QE8xMnqIYidBxG17aJCjTQe7ATrQ5QpZT7U6slvCBcybi8UGqqVRzMCKWwKWapwa+n5\nhcdL4+SIZasnW5IXBayukJtusMEGFfF1teYUHX+7lnmBwrqMBRmrtUlzCPjfPNZ/E0MgawRMIfof\nonD3UJySH5rn5cka7Hr722233bTCNYoJf7UUtHr7LVM7U4iyuZo8pAlOl/peGqy60047hQoR1owb\nbrhBLZS4j2FmR/lBiWk1M3zaiwaWGMa43HLLaXHaI444QhUMlDpf4b3anOKoNcvuHu+n1jrucjBO\nsgLXOtb2/4KAV4h8nOYve2zJEMgAAVxm7ZQsmaqzGjc1wii2KHAGEvcTSIB0Vl031Y+88Sq7MMU1\nq9WeaqrzNh3UaqZqsdAFEv/VptmU8zRiwQlE4Q74TXoRBcMvBrPOOmsgyke4Dqv6qquuGq6zINbN\nljHD07/E2AUSo8eiygknnBCIheV/a4HW/eJ3y28FqTWn8MD/LTTL7h7vRxIm9P4hLrP4rkDCBALJ\nFA2377777onFXcMGtpCKgMSDKs5ieUxtYzsMgWYRMAuR3E3zRGCHjksWm1Qr15gE1k16IjDeeOPV\n5FzqeZRtiSJAkgA8TriccM3CrE62lBfi/fwbOQH2xPFIKRy/Wz+zZoY/+uijNbifZIIkgWtr8 ODB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\n9whwj+JlKI9hDr2fnfXQaQRMIWrxFfAFLWu5C1o8jNTueSuXzB3loGGM WAYIsuVh9cQTT6Qe1+kd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\niLAUYOnZZJNNHBXa8S9Hi/5hOfKcFUnkcj44j5g JLxC9Ibh9vHAe3AW4uuBK8X2S4uuFDBcsTkcd\ndZTGt+BKiwuWIcz9ZJp4wV2Bq0+ULLXK1JqTP85/1hOvUC+5HFawvn37hjER/hx5+8R1Oddcc2md\nrXh9rbyN1cZTiQBcR1A8yN3K7bvvvvqbif6OKlvbmiHwMwKPP/64fmfybLm2a1VsBAqvEJFtArsu\nrircWVSa3nLLLcOr0gy5XJJbwCsUngcjPEFsYfbZZ9ctxOYkKURiMVKFw7vHYofraq05xY+Jph/H\n9zWyjuvw/PPP17TlRo7rVFtujPAnmRQLgbwxwxcLve4dLb/1fv36hcSh3YuEzbxVCBReISLY+fjj\nj9fg35133lkzYwhI5s0TaYZcrlpGWLV9nI8imIiPYdKVyD8CtYlVgrTRK1mR3bpYa07x9lmQyxG3\nRCYP/ChJCmH8nHlYX3755ZU/ibFPPvnkeRhSqceQFdGot8q2GqybbroppA/gXMT1VSv50FtyUqyr\nkEpKnb0eU6vWN8Hhjz32WHgMFuL5558/XLeFnxEgm4/fvIkh0DIEGnay9fKArGOIiM3xcT1ivQkk\nWysQq5COkngiAS4Qcrlw1BROFfdUuH7eeedpmzFjxoTbxK2l24gT8iKB0rpNlBndJMG8uh6PzSGe\nSdKU/WFBPIZICBD1uFNPPTVswwL+cbEa6bZqc6o46H8rSy65ZCCZdVX/mGeagJu4+gKxalU0Efdg\n4OdbsaOOlVbHEDEEUYQCUSoDivOatB6BJ598Mth22231+3vuuee2/oS9PAPxg0OGDAlef/31gN8r\nhV/ThKKsoqgFElyu85MXn+DYY49Na16xXcgX9TfPvUYC/Sv2sVKrb+KohF8nEIoB/T5T/NWkEgHu\nRVwT8QJU7rA1QyBDBApvIcLNAzU+b2ak35KxRYo7EiWX23DDDd2zzz6r5HLEArFPcHSkFSNRcjl/\nHFlMxPYg3lUWJ16DjdqLVG JW7p5o3SLPpuz7hOPnwAMPVDcffeE+oA94YERp0a6qzcmfK/oJuVyz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\nuIgRymtQ1w9GdtzEFI6lBhmUGf3799c28X+9IRyN91VtvZXkpK3su9qcyrR
<TRUNCATED>
