Filtering and Aggregation with Bquery

filter-aggregate.py

import bcolz
import bquery

data_path = '/some/place/with/bcolz/data'

c_table = bquery.ctable(rootdir=data_path)

## Filter

# create the criteria
string_feature = 'some_string_feature'
criterion = "feature == b'{0}'".format(string_feature)

# create the boolean array, with numexpr
boolarr = c_table.eval(criterion)

## Aggregate

# column_name whose unique values define groups
group_columns = ['group_column_name']

# input_column_name, operation, output_column_name
aggregation_operations = [['number_column_name', 'mean', 'mean_column']]

# use boolean array in aggregation
mean_repin_count = c_table.groupby(group_columns,
                                   aggregation_operations, bool_arr=boolarr)