NEURAL NETWORK MODEL FOR PREDICTING THE ABSOLUTE EFFECTIVENESS OF PREVENTIVE MEASURES AND SUPPORTING MANAGEMENT DECISIONS

Nguyen Vuong Anh

The article analyzes and evaluates the effectiveness of preventive measures using neural networks. The input data for training the neural network include information about fires collected in Vietnam, such as population, number of fires, number of deaths, and number of preventive measures. Based on these indicators, the effectiveness of the preventive measures is predicted. The results of the study play a fundamental role in the practices of the General Directorate of the Fire Service in the decision-making process to enhance fire safety in the provinces and cities of Vietnam. The construction of a model for predicting the effectiveness of fire prevention and extinguishment measures in Vietnam is based on a multilayer neural network. Decisions to ensure fire safety in the country can be made based on the forecast results. Efficiency of forecasting and fire risk values have been established for the provinces and cities in Vietnam. These indicators assist in developing models for optimally predicting the effectiveness of preventive measures. The results indicate the feasibility of predicting the absolute value of preventive work's effectiveness using quantitative and categorical variables. A relatively large forecasting error is linked, on one hand, to the necessity of considering a greater number of input parameters, and on the other hand, to the need to expand the data set for training the neural network. After refining the model, the obtained results enable the evaluation of the effectiveness of preventive measures for the provinces and cities in Vietnam.

Keywords:fire safety, preventive measures, neural network, management decision support, prediction model