EcoVision: Machine Learning Driven Air and Water Quality Forecasting System for Sustainable Living

Authors

• Mina Wael, Canadian International College (CIC)
• Amir Rafaat, Canadian International College (CIC)
• Ahmed Tarek, Canadian International College (CIC)
• Khaled Hussein, Canadian International College (CIC)
• Youssed Muhammed, Canadian International College (CIC)
• Mazen Ezzat, Canadian International College (CIC)
• Gehad Ismail Sayed PhD, The British University in EgyptFollow

Document Type

Conference Proceeding

Publication Date

2026

Abstract

Air and water pollution have become a real threat, affecting the environment, human health and economic stability. Both air and water pollution are among the most pressing global environment challenges significantly affecting health issues such as respiratory infections, cardiovascular diseases and waterborne illnesses. This paper offers promising solutions through Ecovision: A machine learning driven air and water quality forecasting system for better sustainable living. The proposed Ecovision system mainly utilizes predictive models for air quality index (AQI) and water quality index (WQI) major pollutants. It consists of two main parts; the front-end and the back-end. In the front-end the user uploads historical data in a csv format. Then, this data is sent to the pre-trained model for predictions. The predictions are then shown to the user with the option to download. The back-end part consists of three phases: data preprocessing, prediction and report, green area calculation. In the preprocessing phase, any unused columns are firstly dropped then timestamp are formatted and missing data are tackled. Then the processed data is used in the predictive models to get predictions. These predictions are then used in calculations to suggest green areas needed. The Ministry of Environment ROC (Taiwan) AQI dataset is used for the air quality pollutants models while the WQI dataset is used for WQI pollutants. The utilized machine leaning algorithms are Support Vector Regression (SVR), Extreme Gradient Boosting (XGBoost) and Long Short-Term Memory (LSTM) are evaluated using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE) and Coefficient of Determination (R²). SVR achieved best results across all air quality pollutants. The proposed model suing SVR achieved RMSE of 0.0398, MAE of 0.327, MSE of 0.016, MAPE of 0.1299 and R2 of 0.9983 for air quality pollutants prediction. Additionally, the proposed model using XGBoost achieved RMSE of 0.0544, MAE of 0.0429, MSE of 0.003, MAPE of 0.2243 and R2 of 0.9967.

Recommended Citation

Wael, M. et al. (2026). EcoVision: Machine Learning Driven Air and Water Quality Forecasting System for Sustainable Living. In: Hassanien, A.E., El-Sayed, E.K., Darwish, A., Snasel, V. (eds) The 9th International Conference on Advanced Machine Learning Technologies and Applications (AMLTA’25), Volume 2. AMLTA 2025. Lecture Notes on Data Engineering and Communications Technologies, vol 274. Springer, Cham. https://doi.org/10.1007/978-3-032-07326-6_12