Refinement of an Environmental Pollution Model for the Needs of the Electric Power Industry by Addition of Precipitation Attributes.

Pollution modeling is a non-trivial task with a strong stochastic character; Previous experiments using satellite and air pollution data have been extended in this paper with rain gauge attributes. This brought a new potential for increasing the accuracy of regression models of pollution, which was used to a certain extent and enabled us to increase the accuracy of the regression model predicting the soluble fraction of pollution deposit. The statistical test verified the significance of the increase in the accuracy of the Random Forest regression model. An interesting result is also an overview of the most significant influences/attributes influencing the modeled quantities of pollution, which can help future experiments in the selection of attributes influencing pollution, subsequent modeling of pollution, and can aid domain experts in the field of environmental science.