Data and Codes for Geospatial Constrained Optimization

The shared files provide data and codes to support the paper, which is titled "Geospatial Constrained Optimization to Simulate and Predict Spatiotemporal Trends of Air Pollutants". <br><br>The library files provde necessary R and Python functions to support the extrapolation and geospatial constrained optimization. Specific explanations are given in these files. <br>Please ensure that the following libraries are installed:<br>R packages: rgdal, sp, raster, spatialEco, SpatioTemporal (https://cran.r-project.org/web/packages/SpatioTemporal/index.html), sptemExp (https://cran.r-project.org/web/packages/sptemExp/index.html).<br>Python packages: pandas, numpy, tensorflow and keras. <br><br>Please take the following steps for the proposed method: <br>step1_basisfuncs.R: this tutorial aims to explain how to generate temporal basis functions based on KNN and geospatial basis functions. <br>step2_basisfuncs_extrapolation.py: this tutorial aims to explain how to use CNN to extrapolate temporal basis functions. <br>step3_constrained_opt.R: this tutorial aims to explain how to use constrained optimization. <br>step4_indepdendenttest_PM25.R: this is to illustrate how to use geospatial constraint optimization for location-based cross validation of PM2.5. <br>step4_indepdendenttest_NO2x.R:this is to illustrate how to use geospatial constraint optimization for location-based cross validation of NO2 and NOx. This also illustrates how to use the other variable (e.g., NOx) to constrain the target variable (NO2). <br><br>