Clustering of household plots using Self-Organizing Maps

Abstract

The article provides a methodology for typing personal subsidiary plots in a settlement in the Astrakhan region of the Russian Federation, implemented through cluster analysis based on the self-organizing map. As a result, 3 clusters were formed: a small cluster of 125 large households, a small cluster of 370 medium-sized households and a large cluster of 1879 small households, the silhouette coefficient, SC value was 0.56. Using the self-organizing map graph, the 3 largest households assigned to cluster C1 were identified. The specialization of cluster C2 for breeding dairy cattle has been established (the share of livestock was 43.9% of the total livestock of the settlement). The specialization of cluster C1 is primarily livestock-raising and consists of cattle breeding (the share is 78.9%), this cluster is also distinguished by the presence of 100% of the settlement’s pastures (all pastures belong to the households of this cluster), the breeding of 98% of goats, 91, 7% sheep. The crop-growing orientation of cluster C1 is manifested in accordance with the share of potato crops, which is 71.4%, and therefore it makes sense to further split cluster C1 to identify large households with crop-growing specialization. Cluster C3 is distinguished by a large area of perennial crops (the share is 79.4%), specialization in breeding pigs (the share in the settlement is 76.6%) and poultry (72.9%).