Thank you for presenting. In the feature selection and extraction part, can you explicitly explain how you chose the threshold of 15 distinct candidates by balancing improved model performance and computational costs? And how the information contained in the excluded donors was retained via feature extraction?
Besides, have you considered letting the algorithms select features by themselves since SVR is good at dealing with high dimensionality and there are many other such algorithms such as Lasso.
Thank you for presenting. In the feature selection and extraction part, can you explicitly explain how you chose the threshold of 15 distinct candidates by balancing improved model performance and computational costs? And how the information contained in the excluded donors was retained via feature extraction? Besides, have you considered letting the algorithms select features by themselves since SVR is good at dealing with high dimensionality and there are many other such algorithms such as Lasso.