No Free Lunch: An ML Models Field Guide with Tezan Sahu

Every model has a bill. Learn what you're paying.

Most ML material sits at one of two extremes: a 5-minute video that makes you feel like you understand, or a 500-page textbook that buries the intuition. Interviews live in the middle, and so does this guide.

No Free Lunch is an interactive web guide to the 20 classical ML models that come up in most recent data-science, ML-engineer and AI-engineer interviews. It's built so you can explain, tune and defend each one, not just recite it.

🍽 Every model page is a nine-course meal

1. The Bill: every assumption and cost the model charges you, as a receipt you can memorise
2. The Setup: a fresh, real-world problem (EV charging, BNPL credit, CI builds, quick-commerce dark stores); no iris, no Titanic
3. Explain it like…: four depths (ELI5, High-schooler, Undergrad, PhD). Start where you're comfortable, climb when you're ready.
4. The Lab: interactive plots. 🌱 Intuitive mode for first-timers, 🔬 Deep-dive mode with every hyperparameter on a slider.
5. The Knobs: what each hyperparameter does, and what happens when you turn it up or down
6. Order it / Skip it: when to use the model, when it's a trap, and what to check in the data first
7. Gotchas: the non-obvious failure modes that sink candidates
8. Interview Room: 10+ questions per model (concept, scenario, debugging, “what if…”), each with a model answer
9. Under the Hood: plain-step pseudocode plus a from-scratch NumPy implementation, the kind interviewers ask you to write, plus the real library call and links to the docs

📚 What's on the menu: 20 models

Staples (must-know):

Linear Regression · Ridge, Lasso & Elastic Net · Logistic Regression · Naive Bayes · k-Nearest Neighbours · Support Vector Machines · Decision Trees · Random Forest · Gradient Boosting (XGBoost, LightGBM, CatBoost) · k-Means · PCA

Specials (good-to-know):

AdaBoost · LDA & QDA · DBSCAN · Hierarchical Clustering · Gaussian Mixture Models · Isolation Forest · t-SNE & UMAP · ARIMA & Exponential Smoothing · Matrix Factorization

🥫 Plus The Pantry:

28 base concepts every model leans on (bias–variance, regularisation, metrics, leakage, calibration, kernels and more), each with a Beginner and a Refresher view and live widgets.

👀 Try before you buy

Linear Regression, Logistic Regression, Decision Trees, k-Means and the first two Pantry shelves are free on the site. If you like those, the other 16 models are the same depth and craft.

👉 https://no-free-lunch.services-tezansahu.workers.dev

🎯 Who it's for

1. Preparing for data-science, ML-engineer or AI-engineer interviews
2. Switching into ML from software, analytics or academia
3. Juniors (0–4 yrs) who can call `.fit()` but want to confidently answer “why this model?”

What you get:

Lifetime access to the full site (all 20 models + all 28 concepts), on any device, including future updates to these pages.