1. Define the question and how success is measured
2. Get and clean data
3. Explore: distributions, correlations, missing values, leakage risks
4. Build a simple baseline first
5. Model, evaluate on held-out data, iterate
6. Deploy, monitor, retrain

Statistics essentials: mean vs median (outliers), variance and standard deviation, correlation is not causation, sampling bias, and the difference between a statistically significant and a practically meaningful effect.