Partition Procedure of the Initial Data for the Models Based on Artificial Neural Networks

A method of the data partitioning to restore the spatial distribution of the trait was proposed. The controlled partitioning procedure of the source data was applied to the model based on artificial neural networks (ANN). The content of chemical element Chromium (Cr) in the surface layer the soil of urbanized area was modelled. The most accurate results were shown by the model, was based on the partitioning method, which took into account the spatial quota and the variation of the values of Cr content in the soil. The described method is relatively simple to implement in modern computational packages and can be used for the modeling of the spatial variables.