How Implementing Machine Learning Ended Up with an If-Else - Lessons from OpenShift Cluster Prediction

Explore a DevConf.CZ 2023 conference talk that delves into the challenges and pivotal decisions in Machine Learning projects. Learn from Juan Díaz and Katya Gordeeva's experiences in extracting value from OpenShift cluster data, including their project on predicting upgrade issues. Gain insights on maintaining focus on business goals, defining success metrics early, and choosing appropriate technologies. Discover the importance of stakeholder communication, addressing moral challenges, and overcoming qualification and scalability issues in ML implementations. Understand why 85% of ML projects fail and how to avoid common pitfalls through practical lessons learned in a real-world scenario.

Intro
Agenda
Context
Problem Statement
Typical Problem
Data Set
Moral Challenges
AI ML
Metric choice
Prediction
False positives
Fscore
Baseline
Iterating
Issues
Theories
Qualification challenges
Scalability
Questions