Determination of rainfall erosivity in the framework of data science using machine learning and geostatistics methods

The subject of the thesis concerns the determination of the rainfall erosivity factor R of the Universal Soil Loss Equation for the whole Greek territory. The methodology falls within the framework of Data Science: the import of recording and non-recording rain gauge time series and a terrain model into a tidy database, after cleaning from errors; the location of erosive storms within the recording gauge time series and the evaluation of their erosivity; data transformation and the computation of new variables and statistics; exploratory data analysis of R and EI; the evaluation of algorithms with cross-validation and inferential statistics for the imputation of EI values and the estimation of R at new locations; and the creation of maps of R.