Lesson 8: Residual Analysis and Forecast Error
October 20, 2025
Review:
- Categorical data and Indicator (aka “dummy”) variables
- Extended multiple regression Quiz next week on Mon, Oct 27
- Used car model with video presentation (see below) due Mon, Nov 3
Presentation:
- Residual Analysis
- Actual vs Predicted
- Demonstrate with Used Car data
- Measurement of forecast error
- MAD – mean absolute deviation
- MAD = ∑|Actual – Forecast|/(Number of Forecasts)
- RMSE – root mean square error
- RMSE = √(∑(Actual – Forecast)^2/(Number of Forecasts))
- Demonstrate
- MAD – mean absolute deviation
Activity:
- Use your Used Car multiple regression model
- Generate residuals
- Create an Actual vs Predicted scatterplot
Assignment:
- Use this more extensive Used Car sales dataset
- Choose a particular Make and Model, e.g., “Honda Fit” or “Ford F-150” (sign up before end of class)
- Build a multiple regression model
- Generate residuals and make an Actual vs Predicted plot
- Produce a video describing your model and showing the Actual vs Predicted plot.
- Submission
- Your model spreadsheet file or link
- Video presentation (1-3 minutes, use YouTube or PowerPoint)
- Send via email to me
- Due before class on Mon, Nov 3
