Application of Decision Trees for Prediction of Sugar Content and Productivity using Soil Properties for Actinidia arguta ‘Autumn Sense’

Abstract

Environmental conditions are important in increasing the fruit sugar content and productivity of the new cultivar Autumn Sense of Actinidia arguta. We analyzed various soil properties at experimental sites in South Korea. A Pearson’s correlation analysis was performed between the soil properties and sugar content or productivity of Autumn Sense. Further, a decision tree was used to determine the optimal soil conditions. The difference in the fruit size, sugar content, and productivity of Autumn Sense across sites was significant, confirming the effects of soil properties. The decision tree analysis showed that a soil C/N ratio of over 11.49 predicted a sugar content of more than 7°Bx at harvest time, and soil electrical capacity below 131.83 μS/cm predicted productivity more than 50 kg/vine at harvest time. Our results present the soil conditions required to increase the sugar content or productivity of Autumn Sense, a new A. arguta cultivar in South Korea.