Scikit‑learn Tutorials

Data Preprocessing with Scikit‑learn

Effective preprocessing is a cornerstone of any Machine Learning pipeline. Scikit‑learn provides a rich set of utilities to transform raw data into a format suitable for modeling.

Why Preprocess?

• Improve model convergence and performance.
• Handle missing values and categorical features.
• Ensure features are on comparable scales.

Common Steps

1. Scaling – StandardScaler, MinMaxScaler, RobustScaler
2. Encoding – OneHotEncoder, OrdinalEncoder
3. Imputation – SimpleImputer, KNNImputer

Code Snippets

StandardScaler

from sklearn.preprocessing import StandardScaler
import pandas as pd

df = pd.read_csv('data.csv')
scaler = StandardScaler()
scaled = scaler.fit_transform(df[['age','salary']])
df[['age_scaled','salary_scaled']] = scaled
print(df.head())

MinMaxScaler

from sklearn.preprocessing import MinMaxScaler

mm = MinMaxScaler()
df[['age_norm','salary_norm']] = mm.fit_transform(df[['age','salary']])

OneHotEncoder

from sklearn.preprocessing import OneHotEncoder

enc = OneHotEncoder(sparse=False, handle_unknown='ignore')
categorical = df[['city']]
encoded = enc.fit_transform(categorical)
encoded_df = pd.DataFrame(encoded, columns=enc.get_feature_names_out(['city']))
df = pd.concat([df, encoded_df], axis=1).drop('city', axis=1)

SimpleImputer

from sklearn.impute import SimpleImputer

imp = SimpleImputer(strategy='median')
df[['age','salary']] = imp.fit_transform(df[['age','salary']])

Interactive Demo – Min‑Max Scaling (JS)