Open AccessSMOOTHING IN NEURAL NETWORK FOR UNIVARIAT TIME SERIES DATA FORECASTING
Author(s) -
Nurfia Oktaviani Syamsiah,
Indah Purwandani
Publication year - 2020
Publication title -
jurnal riset informatika
Language(s) - English
Resource type - Journals
eISSN - 2656-1743
pISSN - 2656-1735
DOI - 10.34288/jri.v3i1.175
Subject(s) - exponential smoothing , artificial neural network , mean squared error , smoothing , data set , computer science , series (stratigraphy) , set (abstract data type) , time series , test set , algorithm , test data , data mining , machine learning , mathematics , artificial intelligence , statistics , paleontology , biology , programming language
Time series data is interesting research material for many people. Not a few models have been produced, but very optimal accuracy has not been obtained. Neural network is one that is widely used because of its ability to understand non-linear relationships between data. This study will combine a neural network with exponential smoothing to produce higher accuracy. Exponential smoothing is one of the best linear methods is used for data set transformation and thereafter the new data set will be used in training and testing the Neural Network model. The resulting model will be evaluated using the standard error measure Root Mean Square Error (RMSE). Each model was compared with its RMSE value and then performed a T-Test. The proposed ES-NN model proved to have better predictive results than using only one method.