Air pollution prediction system using XRSTH-LSTM algorithm

Area of Science:

• Environmental Science
• Computer Science
• Artificial Intelligence

Background:

• Rising global air pollution (AP) poses significant threats to human health and ecosystems.
• Accurate prediction of AP is crucial for developing effective mitigation strategies.
• Existing models often face challenges with computational cost and precision.

Purpose of the Study:

• To develop a novel Artificial Intelligence (AI)-based prediction system for air pollution.
• To enhance the accuracy and precision of air quality index (AQI) prediction.
• To reduce the computational cost associated with AP prediction models.

Main Methods:

• Development of a Xavier Reptile Switan-h-based Long-Short Term Memory (XRSTH-LSTM) model.
• Fine-tuning of the XRSTH-LSTM model through pre-processing, attribute extraction, and AQI prediction.
• Utilizing the Air Quality Data in India (2015-2020) dataset from Kaggle for training and validation.
• Evaluation of model performance using metrics like MSE, MAPE, RMSE, precision, recall, F-measure, negative predicted value, and Mathew correlation coefficient.

Main Results:

• The proposed XRSTH-LSTM model achieved an accuracy of 98.52% and precision of 99.79%.
• This represents a 0.74% improvement in accuracy compared to existing state-of-the-art models.
• The model demonstrated robust performance and efficient processing of air quality data.

Conclusions:

• The novel XRSTH-LSTM model offers a highly accurate and precise solution for air pollution prediction.
• The AI-based approach effectively addresses the need for advanced AP forecasting.
• The developed system shows superior performance over current models in predicting air pollution levels.