Estudio estadístico de fallos en calderas de gas domésticas utilizando diversas plataformas de software

Modern public utilities require a high level of reliability, especially with regard to systems used for heating and hot water supply, such as gas boilers. This paper collected and studied statistical information on the failures of gas boilers of four brands. The data were provided by three service companies that repair gas equipment in Kaliningrad, in the Russian Federation. The specifics of these failures were studied, determining their possible causes, and they were stratified by severity. Service companies typically have to use existing platforms (supplementing them with add-ons) or develop their own software to analyze failure statistics. In this regard, they are interested in the emergence of simple and effective tools for monitoring the quality of gas boiler maintenance and repair work. In this study, we used both the publicly available Scikit-learn library of the Jupyter Notebook environment and a custom program to perform data clustering. The main goal was to conduct a comparative assessment of the reliability of gas boilers of various brands based on the analysis of their failure statistics, as well as to develop a software product that enables such an assessment.