Predicting Financial Manipulation Using an Ensemble-based Approach

Abstract

Financial manipulation becomes a critical issue in corporate transparency due to the increased dependency on stakeholders’ decision-making. The present study proposed a machine learning (ML) driven framework to predict financial manipulation with tertiary classification. The aim is to assess the effectiveness of the Ensemble Bagged Trees (EBT) model in predicting financial manipulation with a greater qualitative hierarchy of financial statements. The supervised ML classification technique is trained and tested using secondary data. The EBT model has provided valuable insight and effectively predicted financial manipulation. The study further enhanced the model using feature selection based on chi-square value and achieved dimensionality reduction using parallel coordination plot analysis. The use of the model may help stakeholders make proper decisions based on public financial information.