Dried Apricot Price Prediction Using Regression and Artificial Neural Networks: The Case of Malatya

Abstract

Türkiye is the global leader in apricot production, with a significant portion of its output concentrated in Malatya. Most of the apricots produced are dried, and approximately 90% of dried apricots are exported. Various studies have been conducted on different aspects of apricots. However, as reviewed in the literature, no study has been found specifically focusing on apricot price forecasting. Yet, predicting future prices can be an important requirement for supporting investors' decision-making processes. The aim of this study is to forecast the future course of dried apricot prices and analyse market dynamics to improve the predictability of the sector. Multiple linear regression was chosen as the method, considering the sensitivity of independent variables to economic fluctuations. Additionally, artificial neural networks were used to process the learning of variables, and the results were compared with regression analysis. Both methods providing successful results in apricot price forecasting. The ranking of the impact of independent variables on apricot prices was found to be similar. The agricultural input price index was identified as the most influential factor affecting apricot prices, followed by inflation and export volume. Given the determining role of the agricultural input price index in apricot prices, measures aimed at reducing production costs could increase producers' profitability. Government-supported price stabilization mechanisms and floor price policies could help balance excessive fluctuations and protect producers. Additionally, increasing bargaining power through producer associations and cooperatives could serve as a stabilizing factor against price volatility.