Solid waste forecasting using modified ANFIS modeling

Area of Science:

• Environmental Science
• Data Science
• Engineering

Background:

• Accurate solid waste generation records are often unavailable in developing countries, hindering effective waste management planning.
• Solid waste generation is influenced by dynamic demographic, economic, and social factors, necessitating sophisticated modeling approaches.
• Existing prediction models face challenges due to data scarcity and the complexity of socio-economic variables.

Purpose of the Study:

• To identify the most influential demographic and economic factors affecting solid waste generation.
• To develop a robust forecasting model for annual solid waste generation using a modified Adaptive Neural Inference System (MANFIS).
• To provide a reliable method for waste prediction in data-scarce developing countries.

Main Methods:

• A systematic approach was employed to determine the key influencing factors on solid waste generation.
• A modified Adaptive Neural Inference System (MANFIS) was developed and optimized for waste forecasting.
• Model performance was rigorously evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R²).

Main Results:

• The most significant input variables identified were specific age groups: 0-14, 15-64, and above 65 years.
• The optimal MANFIS model structure comprised 3 triangular fuzzy membership functions and 27 fuzzy rules.
• The model demonstrated high accuracy, with training RMSE of 0.2678, MAE of 0.045, and R² of 0.99; testing phase yielded RMSE of 3.986, MAE of 0.673, and R² of 0.98.

Conclusions:

• The modified ANFIS model offers a systematic and effective method for predicting annual solid waste generation.
• This approach is particularly valuable for developing countries where reliable waste data is often lacking.
• Accurate solid waste prediction is fundamental for sustainable urban planning and resource management.