python_ml_library_memo.md

Scikit Learn

Useful sources

• https://github.com/ageron/handson-ml

API basics

• methods: fit, transform, predict
• IO: numpy.ndarray with the shape that has sample data number in the first placeholder. ex. (number_of_samples, feature_dim).
  some classes expect 1D array as y. ex. SGDRegressor

Preprocessing

Feature engineering

• Polynomial features

# m = 100
# X.shape == (m, 1)
from sklearn.preprocessing import PolynomialFeatures
poly_features = PolynomialFeatures(degree=2, include_bias=False)
X_poly = poly_features.fit_transform(X)
all(X_poly[:,1] == X_poly[:,0]**2) # True

scaling

• StandardScaler

from sklearn.preprocessing import StandardScaler
std_scaler = StandardScaler()

• MinMaxScaler

models

Linear model

• LinearRegression

from sklearn.linear_model import LinearRegression
lin_reg = LinearRegression()

• Ridge

from sklearn.linear_model import Ridge
ridge_reg = Ridge(alpha)

model evaluation

model selection

• train_test_split

from sklearn.model_selection import train_test_split

metrics

• mean_squared_error

from sklearn.metrics import mean_squared_error
mean_squared_error(y_true, y_pred)