Contents

A working reference across the data science stack. Each topic covers the mechanics, the tradeoffs that matter in practice, and the places where the standard answer breaks down.

Probability & Statistics

Bayes' theorem, distributions, hypothesis testing, A/B test design.

ML Fundamentals

Algorithm mechanics, bias-variance tradeoff, regularization, gradient descent.

Data Collection (SQL)

Window functions, CTEs, query optimization, advanced joins.

Data Processing

Pandas/NumPy vectorization, missing data, feature engineering.

Model Evaluation

Classification and regression metrics, imbalanced data, validation strategies.

Python Coding Patterns

NumPy idioms, comprehensions, collections, and the patterns that come up most.

Experiment Design & Causal Inference

Power analysis, multiple testing, DiD, regression discontinuity.

Time Series Analysis

Stationarity, ARIMA, seasonal decomposition, rolling features for ML.

Business Framing & Communication

Translating model outputs, expected value framework, SCQA, when not to use ML.

ML in Production & Ethics

Model monitoring, drift detection, SHAP explainability, fairness, SR 11-7.

LLMs, RAG, agentic systems, prompt engineering, evaluation, and responsible AI.

TF-IDF, word embeddings, text classification, NER, and generation evaluation (BLEU/ROUGE).

Anomaly Detection

Statistical methods, Isolation Forest, LOF, autoencoders, and evaluation without labels.

Survival Analysis

Censoring, Kaplan-Meier, Cox proportional hazards, and time-to-event modeling for churn and credit.

Advanced Causality

DAGs, backdoor criterion, potential outcomes, double ML, and causal forests for heterogeneous effects.