Objectives:
This paper presents the application of data mining methods in the analysis of traffic accidents in the Republic of Serbia during the period 2022–2025, based on the open dataset available through the Open Data Portal of the Republic of Serbia. The aim of the study is to examine the extent to which spatial, temporal, and structural attributes enable the differentiation between accidents involving material damage only and accidents involving injured or deceased persons.

Material and methods:
The analysis was conducted in the Weka environment using the J48, Random Forest, Naive Bayes, kNN, and Logistic algorithms, with evaluation based on Accuracy, Precision, Recall, F1-score, and ROC AUC measures.

Results:
The results indicate that the Logistic algorithm achieved the best overall performance, while Naive Bayes demonstrated the highest recall for the class of accidents involving injured or deceased persons.

Conclusions:
The results indicate that the Logistic algorithm achieved the best overall performance, while Naive Bayes demonstrated the highest recall for the class of accidents involving injured or deceased persons.