A Hybrid Machine Learning Based Audit Classification: A Meta-Heuristic Approach

Abstract

Audit classification is the process of classifying audits according to their type, goal, or focus areas. In order to manage and report audit activities effectively, it entails classifying audits into various types or categories. The classification facilitates better planning and resource allocation by assisting stakeholders, management, and auditors in comprehending the goals and scope of each audit. The raw audit data is cleaned during the pre-processing stage to remove noise and inconsistencies, and then min-max normalisation and standardisation is applied to ensure robustness and comparability of the data. In order to gather pertinent information and characterise the audit data, the feature extraction step makes use of “statistical measures”. From the extracted data, features are selected through New Hybrid Optimization named- RCSO using Sand Cat Swarm Optimization (SCSO) and Artificial Rabbit Optimization (ARO). From the selected features, classification is processed using Support Vector Machine (SVM) and Optimized Artificial neural networks (ANN). The proposed model is implemented using MATLAB programming. The proposed model can be guaranteed to be a more effective technique than the existing technique in terms of performance metrics because the model's execution is compared to existing technology. The proposed method's performance is compared to that of already-in-use methods like ANN, SVM, KNN, and Naive Bayes. The analysis of the model under consideration includes consideration of its sensitivity, recall, MCC, precision, specificity, F-score, FNR, FPR, NPV, and accuracy.