Fixing Bias in Machine Learning Models (Python + Fairlearn Walkthrough)
In this video, I walk you through a hands-on project on bias mitigation in machine learning models using the famous Adult Income dataset. We’ll explore how bias creeps into ML systems, why overall accuracy can be misleading, and how to fix unfair outcomes with both data-centric and model-centric approaches.
What you’ll learn in this walkthrough:
00:00 – Introduction: Why bias in ML matters
01:38 – Loading & cleaning the Adult Income dataset
11:00 – Baseline Logistic Regression model
13:50 – Subgroup analysis: spotting hidden bias
16:30 – Data-centric fixes: oversampling & targeted balancing
17:40 – Why oversampling can backfire
22:00 – Model-centric fixes with Fairlearn (ThresholdOptimizer)
23:00 – Comparing baseline, balanced, and mitigated models
24:00 – McNemar’s test for statistical significance
25:00 – Key takeaways & conclusion
By the end of this video, you’ll understand:
✔️ How to evaluate subgroup performance, not just overall accuracy
✔️ The pros and cons of oversampling minority groups
✔️ How to use Fairlearn’s ThresholdOptimizer for fairer models
✔️ Why building fair AI is an iterative process
📂 Full code on GitHub: https://github.com/AgnesElza/responsi...
📝 Bias Mitigation Blog Post: https://portfolio.agnesaugustine.com/...
🎓 Responsible AI Course: https://www.cloudskillsboost.google/c...
📊 Math for Data Science Course: https://www.coursera.org/specializati...
👧🏽WHO AM I:
I’m Agnes — a full-time Data Scientist and nano content creator from Kerala, now based in Toronto, Canada. I share content on data science, productivity, self-study tips, and more.
Nesta página do site você pode assistir ao vídeo on-line Fixing Bias in Machine Learning Models (Python + Fairlearn Walkthrough) duração hora minuto segundo em boa qualidade , que foi baixado pelo usuário agnes 27 Agosto 2025, compartilhe o link com seus amigos e conhecidos, no youtube este vídeo já foi visto 153 vezes e gostou 4 espectadores. Boa visualização!
