Hierarchical Cluster Analysis in R for Production and Economic Indicators of the Penitentiary System

Abstract

According to official information from the penitentiary system of the Russian Federation, the structure of the production sector of the penitentiary system of the Russian Federation includes 652 institutions. In 2021, the volume of production of goods and services amounted to 36.8 billion rubles. More than 131 thousand convicts are employed in the institutions of the penitentiary system. The divisions of the Federal Penitentiary Service of Russia are actively organizing work to receive orders for the manufacture of products. Thus, conducting research for the manufacturing sector using modern scientific methods is relevant not only for the penitentiary system, but also for the Russian Federation as a whole. One of these methods is machine learning hierarchical cluster analysis. Its advantages: the ability, regardless of the territories, to consider production and economic indicators of interest and the ability to segment the market with the construction of hierarchies. The purpose of this scientific study is to conduct research in the field of machine learning (hierarchical clustering) for segmenting the production and economic indicators of the penitentiary system. The main tool for implementing hierarchical clustering is the programming language and statistical processing – R (data processing was carried out in the R-Studio environment). The novelty of the work is: the study of production and economic indicators of the penitentiary system, regardless of the territories and the use of relevant machine learning methods for segmentation and division into groups of values of the volume of production of goods. The main scientific results were: the developed algorithm for carrying out hierarchical clustering for the penitentiary system; formed a number of rules and norms for the choice of parameters, data processing, the choice of hyperparameters for hierarchical clustering. In addition, new dependencies were identified for a more global consideration of production and economic indicators.