深青色ppt模版：空间数据回归分析Regression Analysis for Spatial Data【国外优秀研究报告】.ppt

* * * * * * * * * Statistically significant clustering of under and over predictions. Random spatial pattern of under and over predictions. Spatial Autocorrelation * Online help is … helpful! * The 6 checks: Coefficients have the expected sign. No redundancy among explanatory variables. Residuals are normally distributed. Residuals are not spatially autocorrelated. Coefficients are statistically significant. Strong Adjusted R-Squared value. Now are we done? A statistically significant Koenker OLS diagnostic is often evidence that Geographically Weighted Regresion (GWR) will improve model results GWR allows you to explore geographic variation which can help you tailor effective remediation efforts. * Global vs. local regression models OLS Global regression model One equation, calibrated using data from all features Relationships are fixed * Global vs. local regression models OLS Global regression model One equation, calibrated using data from all features Relationships are fixed GWR Local regression model One equation for every feature, calibrated using data from nearby features Relationships are allowed to vary across the study area * For each explanatory variable, GWR creates a coefficient surface showing you where relationships are strongest. GWR Exploring regional variation * Exploratory Regression Analysis Finding a Good Model Exploratory regression sample script Exploratory regression can help you find a properly specified OLS Like stepwise regression… but different The fine print There is a tradeoff Download sample script tools from The Geoprocessing Resource Center: ? Geoprocessing LScott@ESRI.com Resources for learning more…During the conference Technical Workshops Spatial Pattern Analysis ArcGIS Spatial Analyst – Statistical Modeling Geostatistical Analyst Extending ArcGIS with R: Technical Briefing Spatial Analysis Island in the Exhibit Hall Hands-on Learning Lab * - Resources for learning more…After the conference Hot Spot and R