Notes around Machine Learning

Machine Learning Notes

Weight of Evidence: turning each variable into risk prediction

Logistic Regression, Credit Risk

Correlation: one relationship, four different numbers

Correlation

Multicollinearity: when two predictors tell the same story

Linear Regression, Assumptions

Homoscedasticity: when the noise grows with the data

Linear Regression, Assumptions

Probability Calibration: when a good score is not a probability

Calibration, Classification

Class Imbalance: when accuracy lies

Classification, Imbalanced Data