Prediksi Gelombang Corona Dengan Metode Neural Network

Abstract

Until recently the spread of COVID-19 is unstoppable. COVID-19 is caused by RNA Virus that spread widely between humans, mammals, and birds which cause respiratory, enteric, heart and neurologic diseases. Although it is known for respiratory infection, the virus through plasma or serum is also happen often. Therefore, there is still theoretical risk of the virus spreading through blood transfusion. Because there are more cases that shown no symptoms, worries about the spread of COVID-19 is increasing. Several attempts have been done for alleviate mortality rates like mask usage and lockdown quarantine. Neural network adapt on how a human brain works. One of neural network techniques is Multilayer Perceptron (MLP). In MLP, input data is received through one dimension and spread through network until an output is achieved. Every neuron connection on two neighboring layers have one dimensional value that determine the quality of that node. On every input data at each layer calculation is done by the weight of the layer, and then the result will be transformed by using non-linear formula that called as activation function. The result of this research is found by the help of two cross validation technic: GridSearchCV and KFold Cross Validation which gave each 0.943887 and 0.911341 score. The score is achieved using r2 which the best parameter of the model is determined as: relu, 0.1, (10,10), invscaling and lbfgs. Result showns that the proposed model can do the prediction well against the mortality rate of corona.