Product-facing data-science interview on choosing and configuring tree-based ensemble models.

Compare Random Forests with Gradient Boosting Decision Trees such as XGBoost.  When would you prefer one over the other in a production setting? Do tree-based models require feature scaling or normalization?  Explain the theoretical reason and any practical exceptions.

Discuss bias-variance, overfitting control, interpretability, parallelism, sequential boosting, split criteria independence from feature scale.

Tiktok Data Scientist interview question: Choose Between Random Forests and Gradient Boosting Models. {"blocks": [{"key": "fde490c5", "text": "Scenario", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {...