Data Preprocessing

Code

import numpy as np
from sklearn.preprocessing import StandardScaler, MinMaxScaler, OneHotEncoder
from sklearn.impute import SimpleImputer

X = np.array([[1, 2], [3, np.nan], [5, 6], [7, 8]])

# Impute missing values
imputer = SimpleImputer(strategy="mean")
X_imp = imputer.fit_transform(X)

# Standardize (zero mean, unit variance)
scaler = StandardScaler()
X_std = scaler.fit_transform(X_imp)

# Scale to [0, 1]
minmax = MinMaxScaler()
X_mm = minmax.fit_transform(X_imp)

# One-hot encode categorical features
enc = OneHotEncoder(sparse_output=False, handle_unknown="ignore")
X_cat = enc.fit_transform([["red"], ["green"], ["blue"], ["red"]])
print(X_std, X_cat)

Explanation

Preprocessing transformers convert raw data into a numeric matrix suitable for models. StandardScaler centers and scales to unit variance, MinMaxScaler maps to a fixed range, and SimpleImputer fills missing values with a statistic. OneHotEncoder expands categorical columns into binary indicator columns.

More Machine Learning Snippets