Why this decision matters
Pricing, demand forecasting, customer lifetime value, real-estate appraisal, sports analytics — anywhere the question ends in a dollar amount or a quantity, you reach for regression first. It's also the most interpretable model: every coefficient maps directly to a business statement ("each additional year of age is worth $X in spend").
By the end of this topic you'll be able to
Build a simple and a multiple linear regression; interpret coefficients in business language; check the assumptions (linearity, independence, homoscedasticity, normality of residuals); spot multicollinearity; explain R² and RMSE to a non-technical audience.
Materials
Key concepts to know
- Simple linear regression — one predictor, one outcome. The line of best fit.
- Multiple regression — many predictors at once. Each coefficient is the effect holding the others constant.
- R² & adjusted R² — how much of the variance you're explaining.
- RMSE / MAE — how far off your predictions are, in the units of the outcome.
- Multicollinearity — when predictors are too correlated with each other, coefficients become unstable.
- Residual analysis — the leftover errors tell you whether your model is missing something.
- Categorical variables — dummies vs. one-hot encoding; reference categories.
Hands-on demos
Two demo notebooks. The Linear Regression starter walks through the mechanics; the Moneyball demo applies the same techniques to a famous real-world business case.
- Linear Regression Demo (notebook + data)Build a regression from scratch; interpret coefficients; check assumptions.
- Moneyball Demo (notebook + data)The Oakland A's used regression to find undervalued players. Reproduce the analysis.
Practice datasets
Each dataset is paired with a business problem to frame the regression around. Pick whichever decision interests you most.
- Panda Express — Site SelectionPredict revenue at a candidate site given trade-area features.
- Sports — Economic ImpactEstimate the dollar impact of hosting a sporting event.
- Healthcare — Diabetes PreventionPredict A1C reduction from a behavioral intervention.
Practice with games · Pick the right predictive method
Regression is the first predictive technique in the course. These short games help you frame predictive problems and pick the right method — useful here and for every supervised topic that follows.
- Why Predictive Analytics?When prediction beats description, and what makes a problem "predictable."
- Technique PickerMatch the business question to the right technique — regression, classification, clustering, etc.
- Model Selection ChallengePick the best model for each scenario; learn to defend the choice.