Fig. 2: The feature importance based on Boruta feature selection method. | npj Precision Oncology

Fig. 2: The feature importance based on Boruta feature selection method.

From: Development and validation of machine learning models for young-onset colorectal cancer risk stratification

Fig. 2

The horizontal axis presented the name of features, and the vertical axis showed the Z-score of features. The box plot displayed the Z-score of features based on RF algorithm. The green boxes indicated the features confirmed as important, the blue boxes showed the tentative features, and the red boxes represented the features judged as unimportant. Eosinophi Eosinophiles, GGT Gamma glutamyl transpeptidase, Glu Blood glucose, TBil Total bilirubin, TC Total cholesterol, TG Triglycerides, ALP Alkaline phosphatase, PLT Platelets, WBC White blood cells, Mono Monocytes, Lipa Lipoprotein a, RDW Red cell distribution width, A/G Albumin/globulin ratio, ALT Alanine aminotransferase, LDH Lactate dehydrogenase, HDL high-density lipoprotein, Lym Lymphocytes, CEA Carcinoembryonic antigen, Hb Hemoglobin, FIT Fecal immunochemical test, CRP C-reactive protein.

Back to article page