A study on atmospheric pressure in krg using spatial regression (SAR and SEM)

Abstract

The development of using location analysis in all aspects of life encouraged us to use it in this paper for analyzing weather in some Kurdistan regions which are (Sulaimaniyah, Erbil and Duhok) based on the data recorded by the devices placed in 27 different areas which had been recording the climate for all months during 2015. Moran test is used to check spatial dependency in a data and Spatial autoregressive model (SAR), spatial error model (SEM) is created also Lagrange test used to select the significant and alternative spatial model between (SAR and SEM). Different criteria or measures like (R2adj, RMSE, MAPE, AICc) are used for finding the best fit model also for all of them used (rook, bishop and queen) matrixes to specify the spatial correlation. The important result in the practical part shows that SAR model for queen and rook matrixes is significant while SEM model is not significant for all three matrixes and SAR model for queen matrix is best appropriate model.