Skip to main content
. 2020 May 23;9(5):giaa052. doi: 10.1093/gigascience/giaa052

Figure 1:

Figure 1:

High-level scheme of hyperSMURF. Step 1: partitioning of the training set (the minority/positive class is represented in blue, while the majority/negative class is in green). Step 2: application of oversampling and undersampling approaches, and assembling of the training set. Step 3: training of the random forest models. Step 4: testing and aggregation of prediction outcomes.