Author/Authors :
Alpaslan, Faruk Ondokuz Mayıs Üniversitesi, Kurupelit Kampüsü - Fen Edebiyat Fakültesi - İstatistik Bölümü, Turkey , İlter, Damla Ondokuz Mayıs Üniversitesi, Kurupelit Kampüsü - Fen Edebiyat Fakültesi - İstatistik Bölümü, Turkey , Dalar, Ali Zafer Giresun Üniversitesi, Gazipaşa Yerleşkesi - Fen Edebiyat Fakültesi - İstatistik Bölümü, Turkey , Eğrioğlu, Erol Ondokuz Mayıs Üniversitesi, Kurupelit Kampüsü - Fen Edebiyat Fakültesi - İstatistik Bölümü, Turkey
Abstract :
In the literature, various types of artificial neural networks have been used to forecast time series. Artificial neuron model is a mathematical model of the biological neuron. Yadav et al. (2007) proposed a new artificial neuron model that was called as multiplicative neuron model. Multiplicative neuron model artificial neural networks that have got one multiplicative structured neuron have been used to forecast time series. In this study, artificial bee colony algorithm is firstly used to train multiplicative neuron model artificial neural network. The training and testing performances of artificial bee colony algorithm are compared with particle swarm optimization and back propagation algorithm by using two real life time series in an experimental design. Three factors ANOVA and multiple compare tests are used as statistical techniques in this study.
NaturalLanguageKeyword :
Artificial bee colony algorithm , Multiplicative neuron model , Artificial neural networks , Forecasting , Particle swarm optimization , Back propagation algorithm
