In this thesis we gathered multiple years of data regarding Slovenian farmers from different sources. We obtained some personal data, data about their land and land use, animals that they keep and some forms of help they received. We merged the data and created farmer's profiles and aggregated municipalities profiles. We added some demographic attributes to the latter. We visualized information about the land, farmers and municipalities. Three different models were trained on the created profiles: logistic regression, naive Bayes and random forests. We evaluated them with AUC and F1. The best results were obtained with logistic regression but better data is needed. Lastly, we created a nomogram as an example of a tool that can be used by farmers and their advisors.
|
