Forecasting of Emergencies and Incidents on the Territory of the Perm Region Using Machine Learning Methods
Year: 2024
Type: article
Abstract: In the modern world, where the amount of data grows exponentially, machine learning methods are becoming a key tool for predicting emergencies and incidents. Recently, ensemble algorithms have been actively used to solve such problems. One of the most effective approaches is gradient boosting over decision trees, which combines the flexibility of trees and the power of gradient optimization. This article presents an assessment of the possibilities of using the gradient boosting method over decision trees using the example of emergencies and incidents in the Perm Territory. The main advantages and limitations of the method are presented. The conducted studies have proven the effectiveness of the gradient boosting method over decision trees for emergency forecasting tasks. Its ability to take into account complex dependencies and work with heterogeneous data makes it a powerful tool in the arsenal of analysts and forecasting specialists.
Authors: Sergey Lutsenko, Daniil Tashkinov, Vitaly Pleshchev, German Tyurin
Institution: Ministry of Emergency Situations
Language: English
Topic: Safety and Risk Management
Domain: Social Sciences
Open Access status: bronze
OpenAlex ID: W4408167167