[熱身賽第二輪] 數據科學與機器學習(2) [Warm-up Round 2] Data Science and Machine Learning (2)

Big Concept: Data Science and Machine Learning Topic: Overfitting vs. Underfitting: Why the Best Model Is Not Always the Most Complex Bloom's Taxonomy: remember, understand Learning Goal: Deliver a clear lecture explaining the concepts of overfitting and underfitting. Please visually guide the students through the 'training and test error versus model complexity' curves, explicitly identifying where the underfitting region, the optimal 'sweet spot', and the overfitting regions lie on the graph. To ensure the concepts are intuitive, seamlessly integrate a relatable real-world analogy (think an interesting one yourself) alongside a concrete machine learning example. Key Learning Points: Training error vs. test error; generalization; model complexity; bias-variance trade-off (conceptual); learning curves; overfitting signs (train acc ? test acc); underfitting signs (both low)