Study on Air Quality Index Forecasting in Nanjing Based on Time Series Modeling

With the rapid advancement of industrialization and urbanization, air quality has become a global concern. In this paper, GM(1, 1) model, ARIMA model and LSTM model are used to predict the future air quality index in Nanjing. The GM(1, 1) prediction model takes the development coefficient = 0.00012, and the grey role quantity is 1435.236; the LSTM prediction model uses the mean square error (MSE) as the loss function, the Adam optimizer is optimized, the hidden nodes of the hidden layer are taken as 15, and the bath-size is taken as 1. The learning rate is 0.001, and the number of iterations is 300 times, and the ARIMA. The autoregressive order p of the prediction model is taken as 13; the difference order d is taken as 1; and the moving average order is taken as 2. Then the corresponding fitting effects are plotted according to the real and predicted values. Finally, by comparing RMSE, MAE and MAPE, it is concluded that the ARIMA model has better prediction effect. The selection of a suitable prediction model for the future AQI in Nanjing can provide more accurate AQI prediction for Nanjing, which is of great significance for promoting the green and sustainable development of Nanjing.