Skip to main content

Table 3 Classification accuracies observed by 1-holdout cross-validation after permuting diagnoses in our training data set. Results reported as the fraction of samples for which the model’s prediction of the diagnosis matches the label assigned under permutation

From: Statistical classifiers for diagnosing disease from immune repertoires: a case study using multiple sclerosis

Classification Accuracy on the Training Data Set by Exhaustive 1-Holdout Cross-Validation (Labels Assigned under Permutation)
7/23 ≈ 30.4% 11/23 ≈ 47.8% 6/23 ≈ 26.1% 6/23 ≈ 26.1% 7/23 ≈ 30.4%
15/23 ≈ 65.2% 13/23 ≈ 56.5% 11/23 ≈ 47.8% 12/23 ≈ 52.2% 7/23 ≈ 30.4%
18/23 ≈ 78.3% 7/23 ≈ 30.4% 6/23 ≈ 26.1% 8/23 ≈ 34.8% 11/23 ≈ 47.8%
4/23 ≈ 17.4% 11/23 ≈ 47.8% 9/23 ≈ 39.1% 10/23 ≈ 43.5% 7/23 ≈ 30.4%
Average: 40.4%