Regression Techniques to Predict the Growth of Potato Tubers

Abstract

Potatoes (Solanum tuberosum) is one of the highest produced commodities for human consumption, with a global production of 359.07 mega metric ton per year and covering a huge cultivation area in the world. As a result, being able to predict the yield of this crop is an interesting topic, with a high impact on agriculture production. Accordingly, to predict some potato-tuber production and quality indicators, this work innovates by using phenotypical plant characteristics to which soft-computing techniques are applied. More precisely, the Linear Multiple-Regression, the Radial Basis Function Network, the Multilayer Perceptron, and the Support Vector Machine are benchmarked in the present work. Promising results have been obtained, validating the application of soft-computing techniques to predict the growth of potato tubers.