Rank features using logistic regression coefficients
Company: Roblox
Role: Data Scientist
Category: Machine Learning
Difficulty: easy
Interview Round: Take-home Project
You are given a binary classification dataset:
- `X`: a 2D array of shape `(n_samples, n_features)` containing numeric features
- `y`: a 1D binary array of shape `(n_samples,)` with values in {0,1}
- `feature_names`: a list of length `n_features` with the name of each column in `X`
## Task
1. **Normalize** each feature column of `X` using z-score standardization:
\[
X'_{:,j} = \frac{X_{:,j} - \mu_j}{\sigma_j}
\]
where \(\mu_j\) and \(\sigma_j\) are the mean and standard deviation of feature \(j\) computed on the training set.
2. Fit a **logistic regression** model on the normalized features.
3. Rank features by their learned coefficient values (largest to smallest), and return the **top 3 feature names**.
## Output
Return a list of 3 strings: the names of the top-3 features.
## Notes
- Assume binary logistic regression (one coefficient per feature).
- Specify how you handle ties and how you deal with regularization defaults in common libraries.
Quick Answer: This question evaluates understanding of feature scaling, interpretation of logistic regression coefficients as feature importance, and awareness of how regularization and tie-breaking can affect coefficient-based rankings.