Fig. 7
From: Velo-Predictor: an ensemble learning pipeline for RNA velocity prediction
The impact of hyperparameters on DG dataset. a Balance score curve with different hyperparameters. The upper one shows the balance score increases with the increasing top gene number k. After k reaches 20, the balance score begins to oscillate which means 20 is enough for a good performance. The bottom one shows the performance drops when the number of classes exceeds 8. b Confusion matrix of the Stacking model when the number of genes k equals 20 and the number of classes d equals to 8. Each square corresponds to predicted label and a true label, and the value in each square is the number of cells satisfying the condition (color is darker as the number increases). The numbers along the diagnal show the performance of the Stacking model on DG dataset