Forecast of Multiple Weather Indexes Using LSTM Model

In this article, a highly concerning problem, weather prediction, is solved using an LSTM model. Weather prediction plays a crucial role in everyday life. People need to know if it will be rainy in order to participate in daily activities and events. Using the LSTM model to predict the weather is capable of acquiring a higher accuracy than traditional methods. I leveraged the LSTM model and combined multiple weather indices to generate a more accurate weather forecast. Furthermore, by tuning hyperparameters such as learning rate, batch size, and epoch, I reduced the MSE of the temperature prediction of the model to 2.6. Compared with other methods, e.g., Graph convolution, which has a 0.92 R2 score, my model outperforms it with a 0.97 R2 score.