Day 15ML · Linear regression
OLS, normal equations, MLE interpretation
- • Derive the OLS solution θ = (XᵀX)⁻¹Xᵀy. When is XᵀX not invertible?
- • Show that OLS = MLE under Gaussian noise.
Win the fundamentals round
Traditional Machine Learning
Cover supervised learning, feature engineering, model selection, and practical cases that still dominate many MLE screens.
Featured topics
7 topic cards built for interview prep
Each topic includes a summary, practical learning goals, representative interview prompts, and a suggested roadmap day.
Beginner90 minDay 10
Linear Models and Regularization
Prepare regression, logistic regression, loss functions, regularization, optimization, and coefficient interpretation.
Learning objectives
- • Derive the intuition behind MSE, log loss, L1, and L2 regularization
- • Explain coefficient interpretation, multicollinearity, and feature scaling
Intermediate100 minDay 15
Trees, Random Forest, and Boosting
Understand why tree ensembles remain the default baseline for many tabular interview cases.
Learning objectives
- • Contrast bagging and boosting clearly
- • Explain bias-variance trade-offs for trees and ensembles
Intermediate80 minDay 17
SVM, KNN, Naive Bayes, and Classic Baselines
Know the classic algorithms interviewers still use to test assumptions, distance metrics, kernels, and baseline thinking.
Learning objectives
- • Compare margin-based, distance-based, and probabilistic classifiers
- • Explain kernel intuition and why scaling matters for SVM and KNN
Intermediate90 minDay 18
Clustering, PCA, and Anomaly Detection
Prepare for unsupervised ML interview questions where labels are weak or delayed.
Learning objectives
- • Describe clustering choices and what distance metric assumptions imply
- • Explain PCA as compression, denoising, and visualization rather than just a formula
Intermediate95 minDay 19
Feature Engineering and Leakage
Build disciplined intuition around feature creation, train-serving parity, and leakage detection.
Learning objectives
- • Spot target leakage, temporal leakage, and data contamination
- • Choose encoding, scaling, and missing-value strategies pragmatically
Intermediate95 minDay 19
Time Series Forecasting
Add forecasting coverage for trend, seasonality, leakage, backtesting, hierarchical forecasts, and anomaly-aware evaluation.
Learning objectives
- • Explain stationarity, trend, seasonality, lag features, and rolling windows
- • Use time-based validation instead of random splits
Advanced90 minDay 38
Interpretability, Fairness, and Responsible ML
Cover model explanations, fairness metrics, bias audits, and governance trade-offs for sensitive ML products.
Learning objectives
- • Compare global and local explanations such as permutation importance and SHAP
- • Discuss fairness metrics and why they can conflict
Practice prompts
Daily-plan topics tied directly to this pillar
These are pulled from the same 133-day roadmap content used by Browse Questions.
Day 15ML · Linear regression
Assumptions: linearity, independence, homoscedasticity, normality
- • Walk through the four classical assumptions of linear regression and how to diagnose violations.
- • What's heteroscedasticity and how do you fix it?
Day 15ML · Linear regression
Loss functions: MSE, MAE, Huber — when to use each
- • Compare MSE, MAE, and Huber loss — what do you use when outliers matter?
- • Why is Huber loss differentiable and robust at the same time?
Day 15ML · Linear regression
Code linear regression forward pass and training loop
- • Implement a vectorized linear regression forward pass for X @ w + b and state the expected tensor shapes.
- • Implement one gradient-descent training step for linear regression and explain loss vs cost vs prediction error.
Day 16ML · Logistic regression & classification
Sigmoid, log-odds, and the LR loss
- • Why do we use the log-loss (cross-entropy) instead of MSE for logistic regression?
- • Derive the gradient of binary cross-entropy w.r.t. the weights.
Day 16ML · Logistic regression & classification
Precision, recall, F1, ROC-AUC, PR-AUC, calibration
- • When is ROC-AUC misleading? When should you use PR-AUC instead?
- • What is a calibration plot and why does it matter for downstream decisions?
Day 16ML · Logistic regression & classification
Multi-class: one-vs-rest, softmax, multi-label vs multi-class
- • Compare one-vs-rest vs softmax for multi-class classification.
- • What's the difference between multi-class and multi-label, and how does the loss change?
Day 16ML · Logistic regression & classification
Logistic regression interview traps: odds, thresholds, sigmoid vs softmax
- • Why is logistic regression called regression if it solves classification?
- • How do you interpret a logistic-regression coefficient as an odds ratio?