Predicting software stack issues in Tensorflow

Project Thoth is a recommendation engine for Python software stacks. As Python is becoming de-facto language of choice for many data scientists and machine learning engineers, Thoth aggregates information about popular Python machine learning and data science packages. This thesis creates two predictive models capable of using Thoth’s knowledge database to spot patterns causing issues in software stacks and predict which software stacks will likely not work. It includes an overview of Python dependency management environment and an explanation of the dependency resolution problem and its challenges. The results compare the two classification models and present the found patterns for discovering issues.

University

Faculty of Informatics

Date of Completion

spring 2021

Resources

Leader

Tomáš Brázdil

Consultant

Fridolín Pokorný

Student

Tomáš Janíček