Table 1 Details of the best model for each classifier on the training set using a cut-off of 0.1 on the log2 copy ratio. The features (Segments, Size and Chromosomes), considered as covariates in the four linear SVM classifiers, were defined using a cut-off of 0.1 on the log2 copy ratio. A repeated tenfold cross validation with 3 repeats was used to assess the model performance on the training set (n = 117) and meanwhile to select the best cost parameter among 0.01, 0.1, 1, 10, 100. Here we report the details of the best model for each of the four classifiers in terms of cost parameter, accuracy, kappa parameter and training error. 3f: three-feature classifier; 2f: two-feature classifier.

From: Binary classification of copy number alteration profiles in liquid biopsy with potential clinical impact in advanced NSCLC