MLS-C01 Exploratory Data Analysis Practice Question

A data scientist is performing EDA on a dataset of customer churn. The dataset includes a categorical feature 'Region' with 100 unique values. What is the best way to encode this feature for a tree-based model?

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Use the feature as a categorical variable directly in the tree-based model

Many tree-based model implementations (e.g., LightGBM, CatBoost) support categorical features natively, handling high cardinality without encoding. Option A is wrong because frequency encoding can introduce target leakage if applied without proper cross-validation. Option C is wrong because label encoding imposes an ordinal relationship that the tree might misinterpret. Option D is wrong because one-hot encoding with 100 categories creates many sparse columns, leading to inefficiency and potential overfitting.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Replace each category with its frequency in the dataset
Why it's wrong here
Frequency encoding can be useful but may not be optimal for trees; also risks target leakage if not careful.
✓
Use the feature as a categorical variable directly in the tree-based model
Why this is correct
Many tree-based models (e.g., LightGBM, CatBoost) handle high-cardinality categoricals efficiently.
✗
Label encode the feature (assign integers 0-99)
Why it's wrong here
Label encoding implies ordinal relationship, which may mislead tree splits.
✗
One-hot encode the feature
Why it's wrong here
One-hot encoding with 100 categories creates many columns, increasing memory and sparsity.

Go deeper

MLS-C01 Exploratory Data Analysis Practice Question

Why each option matters

Option-by-option breakdown

Related to this question