Interpretable Learning Algorithm Based on XGBoost for Fault Prediction in Optical Network
Chunyu Zhang, Danshi Wang, Chuang Song, Lingling Wang, Jianan Song, Luyao Guan, Min Zhang
Abstract
We propose a fault prediction scheme using interpretable XGBoost based on actual datasets, which not only achieves high accuracy (99.72%) and low positive rate (0.18%), but also reveals the five most remarkable features that caused the fault.
Topics & Concepts
Fault (geology)Computer scienceScheme (mathematics)Artificial intelligenceArtificial neural networkAlgorithmMachine learningData miningPattern recognition (psychology)MathematicsMathematical analysisSeismologyGeologyNeural Networks and Reservoir ComputingRetinal Imaging and AnalysisOptical Network Technologies