ML metrics

classification

• accuracy - % correct
• precision + recall
  • precision - % of correct among retrieved instances
  • recall - # of correct
    • e.g. if you were searching for dog on google, getting back everything would mean 100% recall since you got everything that was relevant, but very low precision
• F1 score
• note confusion matrix -> what kind of accuracy are we getting?

Regression

• mean squared error
• mean absolute error
• R Squared