Predição da germinação de sementes de soja utilizando teste de tetrazólio e inteligência computacional

Abstract

The quality of seeds is crucial for agricultural productivity, and conventional testing methods can be time-consuming and subjective. Therefore, the use of machine learning offers an efficient approach to predicting germination. The objective of this study is to find an algorithm that, together with tetrazolium test data, results in an efficient prediction of soybean seed germination. The following machine learning algorithms were tested: REPTree, M5P Decision Tree, Random Forest, Logistic Regression, Artificial Neural Networks, and Support Vector Machine. The data analysis used correlation coefficient (r) and mean absolute error (MAE) as parameters for algorithm accuracy. The inputs tested were viability data, vigor data, and vigor + viability data. The results highlighted SVM as the most effective algorithm in predicting germination, with the inputs of viability and vigor + viability showing the best results. This study suggests that integrating computational intelligence techniques with the tetrazolium test can make the evaluation of soybean seed quality more efficient, contributing to quick and accurate decision-making in agriculture.