5A2D16 Machine learning project – Overfitting Introduction to AI: ML for Business JAOTOMBO Franck SCHOOTS Vincent (SAINVAL Walkens) (SAVINIEN Jean) Review : the goal of Machine Learning = 1, … , + In essence, machine learning refers to a set of approaches for estimating f. Review : measuring quality of fit MSE = 1 =1 ( ()) 2 We want to choose the method that gives the lowest test MSE, as opposed to the lowest training MSE. When a given method yields a small training MSE but a large test MSE, we are said to be overfitting the data. The Bias-Variance Trade-Off Linear regression provides a very good fit to the data. Linear regression provides a very poor fit to the data. = () + 2 + () The Bias-Variance Trade-Off, in brief Bias-Variance Trade-Off & Regularization Cross-validation 4-fold cross-validation run 1 run 2 run 3 run 4 In practice data TEST SET – set aside first ! SPLIT for TRAIN/VALIDATE or CROSS-VALIDATION
