Study and Implementation of AQI Predictive Recommendation System Based on Artificial Intelligence

Air quality is increasingly becoming polluted in many major cities worldwide, so updating and keeping track of air quality information is gaining more attention. Based on the Air Quality Index (AQI) of Ho Chi Minh City and the hourly variations in pollution from April to June 2023, using time-series analysis to forecast the AQI using Recurrent Neural Network (RNN), Linear Regression (LR), Long Short-Term Memory (LSTM), and Deep Q-Learning algorithms. From there, we developed a mobile application called SmartAQI as a social platform to suggest locations with good air quality by the hour, display the air quality for the next seven days, and identify the positions of SmartAQI destinations in Ho Chi Minh City. We evaluated the machine learning algorithms using RMSE, MAE, and MAPE to predict the AQI. The experimental results of predicting the AQI demonstrated that the RNN model achieved the highest accuracy in this experiment.