В статье представлены результаты разработки и апробации интеллектуальной системы оценки эффективности правоохранительной деятельности территориальных органов МВД России, основанной на интеграции методов многокритериального анализа и машинного обучения. Актуальность исследования обусловлена необходимостью совершенствования инструментария объективной оценки деятельности правоохранительных органов в условиях цифровой трансформации государственного управления. Методологическую основу разработанной системы составляет синтез онтологического инжиниринга, динамической модификации метода TOPSIS и ансамбля прогностических моделей машинного обучения. Эмпирическая валидация проведена на массиве данных пяти региональных управлений МВД России за период 2018-2023 гг., включающем 57 показателей эффективности и 23 фактора внешней среды. Результаты демонстрируют существенное повышение точности оценок (на 17,6% по сравнению с традиционными методиками), снижение дисперсии оценок (коэффициент вариации снизился с 0,31 до 0,18) и улучшение прогностической способности (средняя ошибка прогноза MAPE составила 8,3%). Разработанная интеллектуальная система обеспечивает контекстно-зависимую параметризацию критериев, адаптивное ранжирование подразделений с учетом региональной специфики и проактивную идентификацию проблемных областей. Предложенный подход создает методологический базис для перехода от статичных ретроспективных методик оценивания к динамической многофакторной аналитике эффективности правоохранительной деятельности.
The article presents the results of the development and testing of an intelligent system for evaluating the effectiveness of law enforcement activities of the territorial bodies of the Ministry of Internal Affairs of Russia, based on the integration of multicriteria analysis and machine learning methods. The relevance of the research is determined by the need to improve the tools for an objective assessment of the activities of law enforcement agencies in the context of the digital transformation of public administration. The methodological basis of the developed system is a synthesis of ontological engineering, dynamic modification of the TOPSIS method and an ensemble of predictive machine learning models. Empirical validation was carried out on an array of data from five regional departments of the Ministry of Internal Affairs of Russia for the period 2018-2023, including 57 performance indicators and 23 environmental factors. The results demonstrate a significant increase in the accuracy of estimates (by 17.6% compared to traditional methods), a decrease in the variance of estimates (the coefficient of variation decreased from 0.31 to 0.18) and an improvement in predictive ability (the average error of the MAPE forecast was 8.3%). The developed intelligent system provides context-dependent parameterization of criteria, adaptive ranking of departments taking into account regional specifics and proactive identification of problem areas. The proposed approach creates a methodological basis for the transition from static retrospective assessment methods to dynamic multifactorial analysis of the effectiveness of law enforcement activities.
The article presents the results of the development and testing of an intelligent system for evaluating the effectiveness of law enforcement activities of the territorial bodies of the Ministry of Internal Affairs of Russia, based on the integration of multicriteria analysis and machine learning methods. The relevance of the research is determined by the need to improve the tools for an objective assessment of the activities of law enforcement agencies in the context of the digital transformation of public administration. The methodological basis of the developed system is a synthesis of ontological engineering, dynamic modification of the TOPSIS method and an ensemble of predictive machine learning models. Empirical validation was carried out on an array of data from five regional departments of the Ministry of Internal Affairs of Russia for the period 2018-2023, including 57 performance indicators and 23 environmental factors. The results demonstrate a significant increase in the accuracy of estimates (by 17.6% compared to traditional methods), a decrease in the variance of estimates (the coefficient of variation decreased from 0.31 to 0.18) and an improvement in predictive ability (the average error of the MAPE forecast was 8.3%). The developed intelligent system provides context-dependent parameterization of criteria, adaptive ranking of departments taking into account regional specifics and proactive identification of problem areas. The proposed approach creates a methodological basis for the transition from static retrospective assessment methods to dynamic multifactorial analysis of the effectiveness of law enforcement activities.