Spatialization on grid of meteorological variables with machine learning methods

In this thesis we will develop some machine learning methods in order to interpolate me- teorological information and distribute the knowledge in a grid over Italy. The variables analyzed in this work are the minimum and the maximum temperature and the rain, all at the ground level. To perform the spatialization of these variables we use multi linear regressions with a machine learning approach, Boost Decision Trees and Neural Networks and we compare the results obtained. A common procedure of data preprocessing with this type of variables is also widely covered. After the comparison we show some application of these methods in order to produce operative grids over Italy.