314,552 interview questions from 6,000+ companies.
Explain how supervised and unsupervised learning differ, and ground the distinction in a practical ML example.
Explain the bias-variance tradeoff and how it guides model choice, regularization, and generalization performance.
Approach for handling missing values in a pipeline with data quality checks and repeatable transformations.
Tests your ability to design rigorous experiments aligned to testable hypotheses.
Explain precision, recall, F1-score, and ROC-AUC for a classification model.
Explain how to detect and handle NULL values in SQL using filtering, COALESCE, CASE, and business-aware imputation.
Diagnose why a model is underperforming and decide whether the issue is thresholding, class balance, or a deeper data problem.
Explain common SQL-friendly ways to detect outliers and how to handle them without distorting downstream analysis.
Explain a practical framework for feature engineering, from raw data to validated features that improve generalization.
Build an imbalanced binary classifier for payment fraud detection using cost-sensitive learning, threshold tuning, and precision-recall evaluation.
Compare regularized linear and tree-based models for ad CTR prediction, using bias-variance tradeoffs to guide model selection.
Build an imbalanced binary classifier for card fraud detection using class weighting, resampling, and threshold tuning with PR-focused evaluation.
Explain how to choose between a simpler interpretable model and a more accurate black-box model.
Assess whether a large train-to-validation gap indicates overfitting in an imagery triage classifier and recommend how to validate it.
Determine whether Data Society's course completion model is overfitting by comparing train, validation, and test metrics to a simpler baseline.
Determine whether a patient risk classifier is overfitting when training metrics are strong but validation and holdout performance drop materially.
Build a fraud classifier for a 0.1% positive-rate dataset using imbalance-aware training, threshold tuning, and precision-recall evaluation.
Build an imbalanced binary classifier for Microsoft Store fraud detection using weighted learning, temporal validation, and threshold tuning.
Build an imbalanced binary classifier to predict machinery failure 24 hours ahead using sensor, maintenance, and usage data.
Explain how to evaluate and reason about rare event prediction when the positive class is extremely uncommon.
23 total questions
