Genotype prediction in maize (Zea mays L.) progeny using different predictive models
¹Sumy National Agrarian University, Faculty of engineering and technology, Department of Agroengineering, H. Kondratieva 160, UA40021 Sumy, Ukraine
²Sumy National Agrarian University, Faculty of agrotechnologies and natural resource management, Plant growing Department, H. Kondratieva 160, UA40021 Sumy, Ukraine
³Sumy National Agrarian University, Faculty of agrotechnologies and natural resource management, Department of Physical Education, H. Kondratieva 160, UA40021 Sumy, Ukraine
⁴Sumy National Agrarian University, Faculty of agrotechnologies and natural resource management, Department of Biotechnology and chemistry, H. Kondratieva 160, UA40021 Sumy, Ukraine
*Correspondence: polivanui1@gmail.com
Abstract:
This study utilized two probabilistic methods, Gaussian Naive Bayes (GNB) and Logistic Regression (LR), to predict the genotypes of the offspring of two maize varieties: SC604 and KSC707, based on the phenotypic traits of the parent plant. The predictive performance of both models was evaluated by measuring their overall accuracy and calculating the area under receiver operating characteristic curve (AUC). The overall accuracy of both models ranged from 80% to 89%. The AUC values for the LR models were 0.88 or higher, while the GNB models had AUC values of 0.83 or higher. These results indicated that both models were successful in predicting the genetic makeup of the progeny. Furthermore, it was observed that both models were more accurate in predicting the SC604 genotype, which was found to be more consistent and predictable compared to the KSC707 genotype. A chi-square test was conducted to assess the similarity between the prediction results of the two models, revealing that both models had a similarly high likelihood of making accurate predictions in all scenarios.
Key words:
Gaussian Naive Bayes (GNB), genotype prediction, Logistic Regression (LR), predictive models, Zea mays L.