Machine Learning Zoomcamp - Week 6 - Learning in Public
Once of the central themes in Iain M. Banks' Culture series is that on the one hand, the Culture is not only completely dependent on its AI "Minds", it is also utterly abandoned to their beneficence and sine qua non leadership. Meanwhile on the other hand the enemies of the Culture, the Idirans, the Azadians, the Sichult, etc., are somewhere between allowing AI to fill a minor role, and rejecting AI altogether.
This weeks lesson in the Machine Learning Zoomcamp has taught me (among many other useful things) that the only way I could complete this week's homework, without relying on the advice of Claude, Gemini and Perplexity, would be to quit my day job and devote myself completely to the study of machine learning.
Let's say that again out loud: "The study, of machine learning."
You get my point.
Learnings:
- Decision Trees
- Overfitting of Decision Trees
- Tree branches, leaves, and tree depth
- Decision tree algorithmic equivalent
- Finding best splits
- Revisiting feature importance
- Implementing stopping criteria
- Parameter tuning for decision trees*
- Ensembles and random forest
- Gradient boosting and XGBoost
- The XGBoost tool DMatrix
- Implementing watchlists
- XBoost parameter tuning*
- Comparing and selecting from Decision Tree Classifier, Random Forest Classifier, and XGBoost Method Model
• Parameter tuning is a big one
