machine learning book notes #

types:

• supervised learning (right solutions given)
• unsupervised learning (no solutions given)
• reinforcement learning (reinforce beahviour according to feedback)

supervised learning problems:

• classification (X is a … basend on Y)

• regression (attribute Z of X is … based on >)

• nearest neighbour (to which Y is X most similiar)

• overfitting (model works much better on trainingdata than testdata)

• underfitting (model is too weak to reproduce even trainingdata results)

• choice of features is crucial. less might be better bc less dimensions

• good, clean training data is crucial.