Modelling and ForecastingEnvironmental Protection Expenditures: A Comparative Study of Alternative Methods

Forecasting environmental protection expenditures is important in planning the financing resources needed for policies aimed at environmental protection. In this study, we aimed to determine the highest performing models from the Exponential Smoothing, Box-Jenkins" and Artificial Neural Networks methods and to estimate the environmental protection expenditures of metropolitan municipalities, which have a major role in the realization of environmental protection activities in Türkiye, for the periods from2024:-2025: 4. To evaluate the performance of the models, we used the Mean Absolute Percentage Error (MAPE) criterion and identified the model with the lowest MAPE as the most accurate for forecasting. As a result of the experiments, it was seen that all three models (Seasonal ARIMA, Multiplicative-Seasonal Holt-Winter’s and ANN) produced quite successful results, and the Nonlinear Autoregressive ANN model was more successful in capturing the nonlinear patterns in the data compared to the time series models, albeit with a small difference.