Apply labels to a measures table
INFO:tensorflow:Enabling eager execution
INFO:tensorflow:Enabling v2 tensorshape
INFO:tensorflow:Enabling resource variables
INFO:tensorflow:Enabling tensor equality
INFO:tensorflow:Enabling control flow v2


Overview

A measures table can be labelled so that more advanced analytical methods can be applied, which include between-group tests, classification methods in machine learning, and more. The most used historical labelling heuristic is to split players according to the total amount they wager. The labelling module provides a way to do this for any fractional split (e.g. 80:20, 95:5...) via the label_split method.

Methods

top_split[source]

top_split(measures_table, split_by, percentile=95, loud=False)

Labels player's behavioural measures according to their presence in the top percentile of a given measure (split_by). Label column will be 1 if in the top percentile, 0 if not. E.g. if split_by is 'total_wagered' and percentile is 95, players in the top 5% by total wagered will be labelled 1, and all others 0.

get_labelled_groups[source]

get_labelled_groups(labelled_measures_table, labelname)

Provides a simple way of splitting a labelled measures table into multiple tables each corresponding to a given label.

Descriptive Outputs

Labeled measures tables are a powerful tool in themselves, and can be used in the machine_learning module or used to describe each labeled group.

spending_portions[source]

spending_portions(measures_table)

Computes the percentages of total amount wagered of groups of players by presence in the top percentages of the total amount wagered. E.g. how much does the top 5% of players by total amount wagered account for out of the total amount wagered by everyone. It currently computes this statement for the percentages {1,5,10,25,50}, and prints the statements to the console.