This is the companion repository for the paper titled "Concurrent Spatiotemporal Daily Land Use Regression Modeling and Missing Data Imputation of Fine Particulate Matter Using Distributed Space-Time Expectation Maximization" which is published in Atmospheric Environment journal (http://doi.org/10.1016/j.atmosenv.2019.117202). Currently, we have added the dataset and cleaned code to fit and evaluate the final D-STEM kriging and D-STEM LUR models. Also, the code used for assessment of the imputation models is added.
This folder contains daily PM2.5 concentrations in 2015 and the corresponding pool of 210 potentially predictive variables (PPVs) in 30 monitoring stations in Tehran.
This folder contains the original d-stem software source code, especially its "Src" folder which is required for model building.
This folder contains scripts and functions we are coded to fit D-STEM spatiotemporal kriging and LUR models and also it contains the code developed to assess D-STEM fitted models using various metrics (19 metrics) and from different aspects (spatiotemporal, spatial, and temporal) in h-block cross-validation. Furthermore, the developed code for assessment of the imputation models is added.