Skip to main content

Advanced Machine Learning

Recall that in an ordinary multiple linear regression, we have a set of p predictor variables measuring some response variables to fit a model like:

$$ Y = /deltabeta_0 + /beta_1 X_1 ... _ /beta_n X_n + /epsilon $$

Least Absolute Shrinkage and Selection Operator (LASSO)

Lasso regression is a regularization technique for linear regression models. Regularization is a statistical method to reduce errors caused by overfitting on training data.

The process:

  • Start with full model (all possible features)
  • "Shrink" some coefficients to 0 (exactly)
  • Non-zero coefficents indicate "selected" features

Traditional ridge regression: 

Total Cost = Measure of Fit [AKA RSS(w)] + 

xgboost

SVM