Using Ensembles With Spatial Clustering Approaches Applied in the Delineation of Management Classes in Precision Agriculture

Abstract

This paper describes experiments performed using diff erent approaches for spatial data clustering, aiming to assist the delineation of management classes in Precision Agriculture (PA). These approaches were established from the partitional clustering algorithm Fuzzy c-Means (FCM), traditionally used in PA, and from the hierarchical clustering algorithm HACC-Spatial, especially designed for PA. We also performed experiments using diff erent clustering ensembles approaches, evaluating their behavior to achieve consensus solutions from individual clusterings obtained from attribute splitting or using exclusively FCM or HACC-Spatial. The achieved results exhibited some diff erences between FCM and HACC-Spatial, mainly for the visualization of management classes in the form of maps. The HACCSpatial algorithm achieved, in general, better results when compared to FCM and ensembles approaches. Regarding the consensus clusterings provided by ensembles, we can point out the attempt to achieve agreement results which most closely matches the original clusterings, decreasing or increasing the stratifi cation of the management classes maps.

Data analytics for crop management: a big data view

“…Because of the type of the datasets, which is spatio-temporal, it is not surprising to notice that the majority of the clustering algorithms used are of type partitional. K-means and Fuzzy C-Mean (FCM) are considered among the most popular clustering techniques and heavily used to cluster agricultural data [17,18,84,134,137,142,151,154]. The FCM approach has an advantage over K-means, as it deals better with imprecision and noisy data.…”