Recommender systems based on matrix
factorization for agricultural datasets

Abstract. As hardware devices became increasingly sophisticated and cheaper, and fast and broadband wireless internet connection became available not just in towns but also in remote rural areas, using sensors to collect various kinds of data became common in agriculture and applying these sensors to a wide range of locations using huge databases. However, despite the numerous highly qualified agricultural specialists and this huge amount of data collected, lot of useful information remains hidden. A demand for software naturally arises which is able to handle these enormous databases, and can derive latent information from it to ease decision making in important agricultural situations. We delineate in our paper the effort and outcome while we examined Farm Accountancy Data Network data collected by the Hungarian Research Institute of Agricultural Economics (AKI). Our main interest was to find those questions which can be answered by the use of a matrix factorization (MF) model.