Tutorial 09: Column Transformer · Sharda Learning Center

Level: Intermediate · Part of: ML Learning Roadmap

Case Study: Adult Income Prediction

Scenario

Census data mixes numeric features (age, hours-per-week) with categorical ones (occupation, marital-status). Each type needs different preprocessing in parallel.

Learning Objectives

Build a ColumnTransformer with multiple branches
Apply different transformers per column group
Feed output into a classifier Pipeline
Inspect transformed feature names

Prerequisites

Tutorials 06–08
Python 3.9+, scikit-learn, NumPy, pandas, matplotlib

Dataset

fetch_openml('adult', version=2) — income >50K prediction

Hands-On Solution

Copy and run the complete script below:

import pandas as pd
from sklearn.datasets import fetch_openml
from sklearn.model_selection import train_test_split
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.impute import SimpleImputer
from sklearn.pipeline import Pipeline
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report

adult = fetch_openml("adult", version=2, as_frame=True, parser="auto")
X = adult.data
y = (adult.target == ">50K").astype(int)

num_cols = ["age", "hours-per-week"]
cat_cols = ["workclass", "education", "occupation"]

preprocessor = ColumnTransformer([
    ("num", Pipeline([
        ("imputer", SimpleImputer(strategy="median")),
        ("scaler", StandardScaler()),
    ]), num_cols),
    ("cat", Pipeline([
        ("imputer", SimpleImputer(strategy="most_frequent")),
        ("encoder", OneHotEncoder(handle_unknown="ignore")),
    ]), cat_cols),
])

pipe = Pipeline([
    ("prep", preprocessor),
    ("clf", LogisticRegression(max_iter=1000)),
])

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42, stratify=y)
pipe.fit(X_train, y_train)
print(classification_report(y_test, pipe.predict(X_test), target_names=["<=50K", ">50K"]))

Expected Output

When you run the script, you should see evaluation metrics printed to the console. Some tutorials also save .png plot files in the current directory.

Exercises

Add marital-status to categorical columns.
Use remainder='drop' vs remainder='passthrough' and compare shapes.
Time the pipeline fit. Which step is slowest?

Key Takeaways

ColumnTransformer applies different preprocessing to different columns
Nested Pipelines keep each branch clean and composable
Essential pattern for real-world tabular data

← Tutorial 08: Building scikit-learn Pipelines

Tutorial 10: Polynomial Features for Non-Linear Patterns →

Part of the ML Learning Roadmap — Hands-On with scikit-learn