Comparteix:

Workshop "Bayesian Methodologies for AB Testing", per l'empresa Socialpoint.

28/10/2022

En el marc de l'assignatura "Models Matemàtics de la Tecnologia", el 9 de novembre de 8 a 10 h tindrà lloc aquest workshop obert a tot l'estudiantat de l'FME.

Workshop "Bayesian Methodologies for AB Testing" a càrrec de l'empresa Socialpoint

A banda dels estudiants de l'assignatura MMT, qualsevol altre estudiant de l'FME pot assistir-hi.

El workshop tindrà lloc a l'aula 002.

Abstract

Testing the impact of new features or balancing of existing features is a key component in Product Development in Mobile Games. A/B Testing techniques allow teams to test product changes with a reliable control factor. A/B Testing analysis methods typically use the so-called Frequentist approach to assess differences between test groups. Nonetheless, the results from this approach lack the interpretability that business teams require for decision-making. We are going to present a new approach to A/B Testing analysis, based on the Bayesian approach which offers more business-related metrics for assessing risks and opportunities in decision-making policies.

Agenda
- Problem background
- Hypothesis testing in the Frequentist approach: binomial test
- Bayes theorem
- Beta-binomial model for proportions
- Models for numeric random variables
Software requirements
- Python suite Anaconda.
- Packages: numpy, pandas, scipy, matplotlib (installed in the previous Anaconda suite).