This paper evaluates the performance of three forecasting algorithms, specifically Autoregressive Integrated Moving Average (ARIMA), Seasonal ARIMA (SARIMA), and Exponential Smoothing (ES), in forecasting energy consumption at the electric vehicle (EV) charging stations using real-world data. These complex models are compared with two naive models, which serve as benchmarks for evaluating whether the additional complexity in sophisticated models improves forecasting accuracy. To ensure generalization performance, forecasting algorithms are evaluated using four distinct engagement and location-based datasets. The first dataset consists of 256 charging stations located in Croatia, the second dataset includes 20 charging stations located in Zagreb, the third dataset includes 6 charging stations in Split and the last dataset consists of data recorded on the most visited charging station ’Zagreb-Autobusni’. Finally, all algorithms are compared utilizing three error metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE). Additionally, the naive model’s performance is compared with more complex forecasting algorithms to verify its appropriateness for use as a benchmark model in future research.