Table 5 Classification performance of different approaches in fault detection.

From: Transformer windings defects identification using frequency response analysis and advanced data visualization techniques

Method

precision

recall

F1-score

Accuracy

Remark

FCA

0.988

0.992

0.989

0.989

Best performance across all metrics

FA

0.967

0.967

0.966

0.966

Comparable to PCA, but with better visualization

PCA

0.966

0.966

0.966

0.966

Suitable for high-dimensional data

RF [41]

0.971

0.965

0.963

0.964

Robust but lacks visualization capability

ANN [42]

0.969

0.956

0.959

0.955

Requires complex hyper parameter tuning

GB [43]

0.932

0.932

0.931

0.931

Prone to overfitting on small datasets

DT [44]

0.922

0.909

0.909

0.908

Simple but sensitive to noise