Module 5: Regression Modeling Part 2
November 10, 2024
Review:
- Regression Modeling Part 1
- MLB
- East Side Real Estate
Presentation:
- Video Lecture
- Variable Creation
- Demonstrate with Sample Real Estate data
- Multicollinearity
- Interaction Variables
- combine separate variables mathematically
- add, multiply, etc
- another way to avoid problems with multicollinearity
- Dummy Variables
- incorporate categorical variables
- binary – value is either 1 or 0
- if 0, then the coefficient has no effect on y-hat
- if 1, then the coefficient is added
Assignment:
- Multiple Regression Analysis
- Select the worksheet corresponding to the last digit of your PID from this file: PuebloRealEstate_Exam2Data_ByPID
- Save a copy of the data and convert to .xlsx format
- Use Excel with the Analysis Toolpak Add-in
- Build a regression model to estimate selling price
- You must use 1 interaction variable
- You must use 1 or more Dummy variables
- Experiment with multiple combinations of variables to find the best model, maximizing and balancing R^2 and the F Statistic
- Generate selling price forecasts for the following 4 Pueblo properties (with neighborhood location)
- 4 bed, 3 ba, 2,931 sqft (Belmont)
- 2 bed, 1 ba, 1,650 sqft (Central High School)
- 3 bed, 2 ba, 1391 sqft (Northside/Avenues)
- 5 bed, 3 ba, 3070 sqft (Regency and Meadows)