talperetz

<style>
  [data-custom-class='body'],
  [data-custom-class='body'] * {
    background: transparent !important;
  }

  [data-custom-class='title'],
  [data-custom-class='title'] * {
    font-family: Arial !important;
    font-size: 26px !important;

talperetz

import hyperopt

params_space = {
    'learning_rate': hyperopt.hp.uniform('learning_rate', 0.01, 0.8),
    'max_depth': scope.int(hyperopt.hp.quniform('max_depth', 2, 10, 1)),
    'colsample_bylevel': hyperopt.hp.uniform('colsample_bylevel', 0.5, 1.0),
    'bagging_temperature': hyperopt.hp.uniform('bagging_temperature', 0.0, 100),
    'random_strength': hyperopt.hp.uniform('random_strength', 0.0, 100),
    'scale_pos_weight': hyperopt.hp.uniform('scale_pos_weight', 1.0, 16.0), # change 16.0 to n_negative / n_poistive
}

talperetz

from skopt.space import Real, Integer

params_space = [Real(0.01, 0.8, name='learning_rate'), 
                Integer(2, 10, name='max_depth'), 
                Real(0.5, 1.0, name='colsample_bylevel'), 
                Real(1.0, 16.0, name='scale_pos_weight'), # change 16.0 to n_negative / n_poistive
                Real(0.0, 100, name='bagging_temperature'), 
                Real(0.0, 100, name='random_strength'), 
                Real(0.0, 100, name='reg_lambda')]

talperetz

talperetz

# Requirements: script should be ran on relevant mongo instance
# Input: <google storage path> <db_name> <collection_name>
# TL;DR: loads multiple csv files in to mongo collection
# Author: Tal Peretz

# init vars
gs_path=$1
db_name=$2
collection_name=$3
temp_csv_name=partial_csv