Figure 4
From: An accurate and interpretable model for siRNA efficacy prediction
Importance of the number of features. This curve shows the accuracy obtained by the LASSO in a cross-validation as a function of the L1 norm of the weight vector. Small L1 norms generally correspond to sparse models involving a small number of features. The fact that this curve reaches its maximum for large L1 norms shows that most if not all features contribute to the accuracy of the model.