Forecasting the Direction of Agricultural Commodity Price Index through ANN, SVM and Decision Tree: Evidence from Raisin

Akin B., Dizbay I. E., Gumusoglu S., Guducu E.

EGE ACADEMIC REVIEW, vol.18, no.4, pp.579-589, 2018 (ESCI) identifier

  • Publication Type: Article / Article
  • Volume: 18 Issue: 4
  • Publication Date: 2018
  • Doi Number: 10.21121/eab.2018442988
  • Journal Name: EGE ACADEMIC REVIEW
  • Journal Indexes: Emerging Sources Citation Index (ESCI), TR DİZİN (ULAKBİM)
  • Page Numbers: pp.579-589
  • Keywords: Commodity market, Artificial neural networks, Decision tree, Support vector machines, Social & political issues, NEURAL-NETWORK, TIME-SERIES, OIL, GOLD, PREDICTION, MARKET, FOOD, US
  • Dokuz Eylül University Affiliated: No

Abstract

To be able to make appropriate actions during buying, selling or holding decisions, economic actors need accurate commodity price forecasts. This study focuses on forecasting raisin price by using predetermined volatile variables. Therefore, we seek for answers of three main questions. Do the social & political issues effect raisin price in countries that have internal disturbance? By using volatile variables, can we represent or predict price index thoroughly? Lastly, which method has the best prediction performance; Artificial Neural Networks (ANN), Decision Tree or Support Vector Machine (SVM)? In accordance with these purposes, ANN, decision tree and SVM methods are implemented for proposed model and their prediction performances are compared. Experimental results showed that accuracy performance of SVM method was found significantly better than ANN method and decision tree.