Geographically Weighted Regression (GWR)
Feature Description
Geographically Weighted Regression Analysis
Parameter Description
| Parameter | Default Value | Description | Parameter Type |
|---|---|---|---|
| Source Dataset | Specifies the dataset to be calculated. It can be a point, line, or polygon dataset | DatasetVector | |
| Target Datasource | Specifies the datasource storing the result dataset | Datasource | |
| Result Dataset Name | Specifies the name of the result dataset | String | |
| Bandwidth Determination Method | AICC | Method for determining bandwidth | BandWidthType |
| Explanatory Field Name | Name of the explanatory field | String | |
| Kernel Function Type | GAUSSIAN | Type of kernel function | KernelFunction |
| Modeling Field | Field for modeling | String | |
| Kernel Type | FIXED | Type of kernel | KernelType |
| Neighbor Count (Optional) |
0 | Number of neighbors | Integer |
| Bandwidth Range (Optional) |
0.0 | Range of bandwidth | Double |
| GWR Prediction Dataset (Optional) |
Dataset for geographically weighted regression prediction | DatasetVector | |
| Prediction Result Datasource (Optional) |
Datasource for storing prediction results | Datasource | |
| Prediction Result Dataset Name (Optional) |
Name of the prediction result dataset | String | |
| Field Mapping for Prediction Data (Optional) |
Field mapping for prediction data | Object |
Output
Analysis Results: Include geographically weighted regression summary results and result dataset.
The result dataset contains the following attribute fields:
Cross-Validation (CVScore), Predicted Value (Predicted), Regression Coefficients (Intercept, C1_ExplanatoryField), Residual (Residual), Standard Error (StdError), Coefficient Standard Errors (SE_Intercept, SE1_ExplanatoryField), Pseudo t-values (TV_Intercept, TV1_ExplanatoryField), and Studentised Residual (StdResidual). As shown below:
- Cross-Validation (CVScore): The squared sum of differences between estimated and actual values during cross-validation. Serves as a model performance metric.
- Predicted Value (Predicted): Estimated values obtained from geographically weighted regression.
- Regression Coefficient (Intercept): Intercept term of the regression model, representing the predicted value when all explanatory variables are zero.
- Regression Coefficient (C1_ExplanatoryField): Indicates the strength and direction of relationship between explanatory and dependent variables. Positive values denote positive correlation, negative values indicate inverse correlation.
- Residual (Residual): Unexplained portion of the dependent variable (difference between predicted and actual values). Smaller residuals indicate better model fit.
- Standard Error (StdError): Measures reliability of estimates. Smaller values suggest better model performance.
- Coefficient Standard Error (SE_Intercept, SE1_ExplanatoryField): Assesses reliability of coefficient estimates. Larger values may indicate local multicollinearity issues.
- Pseudo t-value (TV_Intercept, TV1_ExplanatoryField): Tests coefficient significance. Values exceeding critical thresholds indicate statistically significant coefficients.
- Studentised Residual (StdResidual): Residual-to-standard-error ratio. Values within (-2, 2) suggest normal distribution and homoscedasticity; values outside this range indicate outliers.