Prediction analysis of COVID-19 in Selangor by using Backpropagation Algorithm with Conjugate Gradient Method

Abstract

COVID-19 is a disease that can spread rapidly among individuals in a population. COVID-19 is a kind of coronavirus. Patients infected with COVID-19 can lead to death, especially people with lung difficulties or a weakened immune system. COVID-19 is transmitted to humans through contact, coughing, sneezing, or close contact. As a result, using previous COVID-19 data in Selangor, an artificial neural network (ANN) is used as an effective future prediction method. Backpropagation is a form of artificial neural network (ANN) algorithm that may be used to resolve issues in prediction analysis. Predictive analysis is utilized to take some earlier action to produce more effective outcomes, which is minimizing the number of COVID-19 patients, to avoid the COVID-19 disease spreading drastically and becoming worse. However, total performance is determined by the optimization approach discovered throughout the training phase. The Fletcher-Reeves approach can improve the efficiency of the backpropagation algorithm by having a faster convergence rate than other methods such as the scaled conjugate gradient method. Based on this theory, this study developed a backpropagation neural network using the conjugate gradient method to create the prediction analysis of COVID-19 patients in Selangor. The result of the experiment will show how many COVID-19 patients can be obtained from the mean square error (MSE) of the test data at various learning rates. The effective learning rate is determined by applying the MSE to both the test and training data, which results in the lowest number of MSE, which is the ideal option. From the output of testing data, the number of predicted confirmed COVID-19 patients can be determined.