Table 10 Comparison of STP and TL prediction accuracy when TL was initialized with correct ATC codes
From: Predicting anatomic therapeutic chemical classification codes using tiered learning
Comparison of N(k) of STP and TL when compared on Supervised learning initialization dataset | ||||||
|---|---|---|---|---|---|---|
|  | STP N(2) | TL N(2) (+/−) | STP N(3) | TL N(3) (+/−) | STP N(4) | TL N(4) (+/−) |
MLP | 85.40% | 87.35% (1.95%) | 77.62% | 77.62% (0.00%) | 52.07% | 51.34% (−.73%) |
Naive Bayes | 57.69% | 78.85% (21.15%) | 48.85% | 66.54% (17.69%) | 31.92% | 41.54% (9.62%) |
Random Forest | 81.03% | 87.26% (6.23%) | 70.46% | 72.63% (2.17%) | 39.84% | 45.26% (5.42%) |
SVM | 83.29% | 88.03% (4.74%) | 76.06% | 79.30% (3.24%) | 53.87% | 56.61% (2.74%) |