Table 16 Statistical analysis of model performance.
From: Automatic smart brain tumor classification and prediction system using deep learning
Model Comparison | Cohen’s d | Confidence Interval | t-value | p-value | Interpretation |
|---|---|---|---|---|---|
Inception vs. EfficientNet-B4 | 1.7941 | Inception: (98.78% to 99.82%) EfficientNet-B4: (99.36% to 100.04%) | \(-38.9958\) | \(7.87 \times 10^{-248}\) | Large effect size; EfficientNet-B4 performs better. |
Inception vs. Custom CNN | 6.2049 | Inception: (98.78% to 99.82%) CNN: (95.36% to 97.64%) | \(138.9141\) | \(10^{-308}\) | Extremely large effect size; Inception performs better. |
EfficientNet-B4 vs. Custom CNN | 7.4635 | Inception: (98.78% to 99.82%) EfficientNet-B4: (99.36% to 100.04%) | \(-38.9958\) | \(< 10^{-308}\) | Large effect size; EfficientNet-B4 performs better. |
