Table 3 Differences among classifier algorithms for staged and spiral processing
From: Systematic review using a spiral approach with machine learning
Staged | Spiral | |||||
|---|---|---|---|---|---|---|
Cultural | PTSD | CBT | Cultural | PTSD | CBT | |
Random forest | ||||||
WSS95% | 6852 | 4552 | 4752 | 5502 | 702 | 7752 |
% of articles | 76.50% | 74.00% | 43.60% | 61.50% | 11% | 71.10% |
Precision | 19.12% | 0.90% | 1.46% | 23.81% | 5.82% | 0.89% |
Accuracy | 48.49% | 26.64% | 57.14% | 60.89% | 89.22% | 29.7% |
Seconds required | 116.6 | 62.4 | 78.5 | 1763 | 7.87 | 202.2 |
Logistic regression | ||||||
WSS95% | 6802 | 4402 | 4952 | 5802 | 452 | 4252 |
% of articles | 76% | 71.60% | 45.40% | 64.80% | 7% | 39.00% |
Precision | 19.26% | 0.93% | 1.40% | 22.58% | 9.04% | 1.63% |
Accuracy | 48.95% | 29.08% | 55.31% | 58.14% | 93.28% | 61.71% |
Seconds required | 184.1 | 83.5 | 124 | 1841 | 23.4 | 155.9 |
SVM | ||||||
WSS95% | 7052 | 4752 | 5502 | 6102 | 952 | 6602 |
% of articles | 79% | 77.20% | 50.50% | 68.20% | 15.50% | 60.60% |
Precision | 18.58% | 0.86% | 1.26% | 21.47% | 4.29% | 1.05% |
Accuracy | 46.66% | 23.39% | 50.28% | 55.38% | 85.16% | 40.22% |
Seconds required | 513.3 | 140.7 | 289 | 5750 | 11.78 | 300.5 |
Naïve Bayes | ||||||
WSS95% | 8102 | 4502 | 5652 | 7052 | 1952 | 7202 |
% of articles | 91% | 73.20% | 51.80% | 78.80% | 31.70% | 66.10% |
Precision | 16.17% | 0.91% | 1.23% | 18.58% | 2.09% | 0.96% |
Accuracy | 37.01% | 27.45% | 48.91% | 46.66% | 68.90% | 34.73% |
Seconds required | 73.23 | 39.87 | 82.4 | 1668 | 37.55 | 187.5 |