Weather predictions from combined parameters with artificial neural network algorithm based on multi-sensor weather data

Abstract

Weather prediction is a scientific process that predicts the state of atmosphere at a certain location and time frame. Weather prediction is an important thing for certain activities because weather condition gives limitations on some activities that can be done. Predictions can be made in various ways. One of them is by using the Artificial Neural Networks (ANN) on multi-sensor weather data. The data used includes various parameters such as temperature, humidity, precipitation, solar irradiance, and wind velocity, collected from a multi-sensor network. In this paper, a weather prediction model was developed using the ANN algorithm, consisting of four layers: an input layer, two hidden layers, and an output layer. Testing was conducted with various proportion of training/testing data at 90%/10%, 80%/20%, and 70%/30%, each at 100 and 150 epochs. The model's performance was evaluated using the metrics of accuracy and Root Mean Squared Error (RMSE). The study results indicate that the ANN model predicts weather parameters with a high level of accuracy in the testing scenario using 150 epochs with a 70/30 data split. This research proves that with simple ANN model, the Indonesian weather that mostly miss-forecasted can be accurately predicted.