Tutorial 06: Categorical Encoding

Level: Beginner · Part of: ML Learning Roadmap

← Back to roadmap

Case Study: Employee Attrition Prediction

Scenario

HR data contains categorical fields (department, job role, education). ML algorithms need numeric representations.

Learning Objectives

• Use OneHotEncoder for nominal categories
• Use OrdinalEncoder for ordered categories
• Handle unknown categories at inference time
• Combine encoded features with numeric columns

Prerequisites

• Tutorial 05
• Python 3.9+, scikit-learn, NumPy, pandas, matplotlib

Dataset

Synthetic employee dataset (generated in code)

Hands-On Solution

Copy and run the complete script below:

import pandas as pd
import numpy as np
from sklearn.preprocessing import OneHotEncoder, OrdinalEncoder
from sklearn.compose import ColumnTransformer

np.random.seed(42)
n = 200
df = pd.DataFrame({
    "age": np.random.randint(22, 60, n),
    "salary": np.random.randint(30000, 120000, n),
    "department": np.random.choice(["HR", "Engineering", "Sales"], n),
    "education": np.random.choice(["Bachelor", "Master", "PhD"], n),
    "left_company": np.random.randint(0, 2, n),
})

X = df.drop("left_company", axis=1)
y = df["left_company"]

preprocessor = ColumnTransformer([
    ("num", "passthrough", ["age", "salary"]),
    ("dept", OneHotEncoder(handle_unknown="ignore"), ["department"]),
    ("edu", OrdinalEncoder(categories=[["Bachelor", "Master", "PhD"]]), ["education"]),
])

X_encoded = preprocessor.fit_transform(X)
print(f"Original shape: {X.shape}")
print(f"Encoded shape:  {X_encoded.shape}")
print(f"Feature names: {preprocessor.get_feature_names_out()}")

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

1. Add a new department ‘Legal’ in test data. Verify handle_unknown='ignore' works.
2. Compare OneHotEncoder output with pd.get_dummies.
3. Why is OrdinalEncoder risky for unordered categories like department?

Key Takeaways

• One-hot encoding for nominal categories without order
• Ordinal encoding only when a true order exists
• ColumnTransformer cleanly handles mixed feature types

Navigation

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