Machine Learning Practice

TDDE01 - Maskininlärning | Linköpings university

Lab 1 contains

• Spam filter done with k-nearest neighbor
• Maximum likelihood
• Feature selection in linear regression by using own implementet k-fold CV
• Polynomial regression models
• Akaike information criterion
• Ridge regression
• LASSO regression

Lab 2 contains

• Decision trees (Gini and Deviance) to predict whether or not a new customer is likely to pay back the loan.
• Regression trees
• Confidence and prediction bands
• Principle Component Analysis (PCA) to find the features that explains the data most
• Independent Component Analysis (ICA)

Lab 3 contains

• Sum of three gaussian kernels to forcast temperatures for a date and place in Sweden
• Train a neural network to learn the trigonometric sine function