Conformal Anomaly Detection based on Association Rules

Ilia Nouretdinov, James Gammerman, Daljit Rehal

Research output: Other contribution

73 Downloads (Pure)


• We propose a novel technique based on a combination of association rule learning and conformal prediction in its Mondrian form.
• As an application, we use data about (anonymised) business customers of a multinational energy company, Centrica plc.
• There are multiple fields in Centrica's SAP database indicating if a customer is an Industrial Corporation or Small/Medium-sized Enterprise. We consider these as labels.
• Often these labels are incorrect or inconsistent across the SAP system, which has afinancialimpacton thecompany. Theaimof this work is to use machine learning to identify potential errors and propose corrections.
Original languageEnglish
Media of outputposter presentation
PublisherCOPA 2019 : 8th Symposium on Conformal and Probabilistic Prediction with Applications
Number of pages1
Publication statusPublished - 10 Sept 2019

Cite this