Research Paper Volume 16, Issue 16 pp 11824—11842
Identifying novel circadian rhythm biomarkers for diagnosis and prognosis of melanoma by an integrated bioinformatics and machine learning approach
Figure 4. Identification of melanoma-related key CRGs by machine learning approaches. (A) Feature CRG selection using SVM-RFE algorithm. (B) Importance ranking of CRGs using a random forest algorithm. The top 20 CRGs ranked by importance were selected as feature genes. (C) Selection of melanoma-associated feature genes using LASSO regression model. (D) Identification of melanoma-related key CRGs. The overlapping feature genes (CRGs) from the three machine learning approaches were defined as melanoma-related key CRGs.