Fig. 2
From: Effect of missing data on multitask prediction methods
Results of training data sparseness on HTSFP subsets. a Median MCC values for DNN and Macau on the HTSFP5 subset. b The median MCC values for HTSFP5 are scaled relative to the performance of the model on the full training set. c Median MCC values on the HTSFP10 subset. d The median MCC values for HTSFP10 are scaled relative to the model with full training set. The color scheme follows the description in Fig. 1