Growth and Yield of Tomato (Solanum lycopersicum L.) as Affected by Hydroponics, Greenhouse and Irrigation Regimes

Abstract

Predictive model of the crop yields provide a framework for agricultural decision making and understanding how different features affect yield. Among various types of machine learning algorithms, Support Vector Machines (SVM) stand out due to ability to generalize well even with limited training samples. The SVM regression approach proposed in this paper outperforms our previous results obtained with model trees. However, SVM regression as a black-box model lacks explanation of prediction. To overcome this limits, we extended our method with the latest algorithms for explaining regression models through analysis of features contributions. This enables the domain experts to interpret and evaluate discovered knowledge.