The paper presents the application of predictive analytics in analyzing traffic accidents in Belgrade for the period 2020–2025. The dataset, obtained from the official Open Data Portal of the Republic of Serbia, was processed using Python and Power BI. Through data cleaning, modeling, and visualization, regression, classification, and clustering models were developed to predict high-risk periods and locations. Results are displayed in an interactive Power BI dashboard showing temporal trends, geographic distributions, and accident forecasts. The analysis reveals seasonal and spatial patterns that can support improvements in urban traffic safety.