COMPARISON OF PERFORMANCE OF DIFFERENT K VALUES WITH K-FOLD CROSS VALIDATION IN A GRAPH-BASED LEARNING MODEL FOR IncRNA-DISEASE PREDICTION

Abstract

In machine learning, the k value in the k-fold cross-validation method significantly affects the performance of the created model. In the studies that have been done, the k value is usually taken as five or ten because these two values are thought to produce average estimates. However, there is no official rule. It has been observed that few studies have been carried out to use different k values in the training of different models. In this study, a performance evaluation was performed on the IncRNA-disease model using various k values (2, 3, 4, 5, 6, 7, 8, 9, and 10) and datasets. The obtained results were compared and the most suitable k value for the model was determined. In future studies, it is aimed to carry out a more comprehensive study by increasing the number of data sets.