The article outlines a protocol for mapping soil in São Paulo Central West, Brazil using existing data and magnetic signatures. The researchers collected soil samples and used a classification analysis to create soil mapping units. They then used soil attributes to group the soil into clusters. The results showed that the Support Vector Machines algorithm performed well in classifying both clusters and soil mapping units. The distribution of soil attributes within each cluster was more consistent than within the mapping units, suggesting that numerical classification can be effective for soil mapping in tropical regions.