Interpretation and Understanding Outputs

Notebook Outline:

An example of hedonic house price modeling using MGWR

• Global Model Results
• MGWR Results - Model Fit
• MGWR Results - Bandwidth Parameters
• MGWR Results - Parameter Estimates
  
  

Back to the main page

Global Model Results

---

Global Regression Model Fit and Parameter Estimates

Two files are saved on your computer once you run the MGWR model - MGWR_session_summary.txt and MGWR_session_results.csv

MGWR Results - Model Fit

MGWR Results - Bandwidth Parameters

Total number of census tracts = 357

Visualizing bandwidths to understand scale

MGWR Results - Parameter Estimates

The new columns added to the results csv:

1. ols_residual - Residuals from OLS model for every location point
2. mgwr_yhat - yhat from MGWR for each location point
3. mgwr_residual - Residuals from MGWR for every location point
4. localR2 - Explained variation at each location point
5. beta_ - Columns starting with beta are parameter estimates for the covariates
6. se_ - Columns starting with se are standard errors for estimates for the covariates
7. t_ - Columns starting with t are t values for estimates for the covariates
8. p_ - Columns starting with p are p values for estimates for the covariates
9. sumW_ - Sum of weights for the covariates at each location

The beta_ columns need to be filtered to have only the significant parameter estimates. This filtering is done using the t_ value columns. The thresholds for t values are defined by the adjusted t values as shown below.

Next

Previous

Back to the main page

---