DATA100-L13: Gradient Descent, Feature Engineering
Contents
$\downarrow{shuffle}$
SGD: Stochastic Gradient Descent(but size == 1)
convexity (凹凸性)
feature engineering
在于怎么使用transforming
feature function
see website code
non-numeric features
one-hot encoding
concat
high order polynomials
detect overfitting
- collect more data
- more see next lecture