Explore machine learning applications in continuous integration through this conference talk from ODSC Europe 2019. Discover how Andrea Frittoli from IBM and Kyra Wulffert from Hewlett Packard leverage open-source tools like TensorFlow and Pandas to extract valuable insights from CI/CD pipeline data. Learn about their experience training ML models using OpenStack's CI dataset and implementing a Kubernetes application for automated failure identification and analysis. Gain insights into binary classification and multi-class problems, data collection, normalization, and experiment workflows. Understand how these techniques can be applied to any CI system to improve DevOps processes and automate test result predictions.
Machine Learning for Continuous Integration - Andrea Frittoli & Kyra Wulffert - ODSC Europe 2019
Overview
Syllabus
Intro
The Team
The OpenStack use case
Collecting data
Data Selection
Data Normalization
Building the dataset
Experiment Workflow
Training Infrastructure
Prediction
Changing test job
Binary Classification - Summary
Multi Class - Changing Resolution
Multi Class - Network topology
Multi Class - Tuning network topology
Multi Class. Summary
Conclusions
Taught by
Open Data Science
