What is it about?
It is shown that the classification results are sensitive to the period of the user inactivity. Too small period leads to inaccurate definition of the churner, creating a lot of false positives in the data labelling, i.e. the labels themselves are wrong and serve as a source of errors.
Featured Image
Why is it important?
From practical point of view the rules for labelling the data are essential part of methodology. If the labelling is done in the wrong way it cripples the results making the practical usage of developed prediction questionable.
Perspectives
Scientists used to concentrate on the classification accuracy too much, the Machine learning discipline should include all steps of modelling to analyze the impact of the solutions on the actual practical measurements, rather than classification technical metrics.
Andrej Bugajev
Vilnius Gediminas Technical University
Read the Original
This page is a summary of: The Impact of Churn Labelling Rules on Churn Prediction in Telecommunications, Informatica, January 2022, Vilnius University Press,
DOI: 10.15388/22-infor484.
You can read the full text:
Contributors
The following have contributed to this page
