Build Classifier with Many Covariates

HardSystem Design00:00

Practice interviewer

In session

5 left

00:00

Your question is Build Classifier with Many Covariates. Take a moment with it on the right.

Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).

You need to log in / sign up to chat or submit.

Problem

Build a classifier on a dataset with 6 core features plus 294 additional covariates

Asked in the OA stage. Q3 was described by a commenter asking what the 294 covariates are for. Explain how you would develop, validate, deploy, and monitor the classifier without assuming a specific business domain.

Clarify the target, prediction-time availability, label timing, and class balance.
Design preprocessing and feature-selection or regularization strategies.
Choose and justify a baseline and production model.
Describe batch or online serving, reproducibility, and training-serving consistency.
Define offline and production evaluation, monitoring, and failure mitigations.