DescriptionThe traditional audit is retroactive in nature and requires some time to process and is subject to substantial latency. With the evolution of technology, assurance processes could be automated and accelerated to provide more frequent and may be preventive audits. This study contributes to the assurance literature by proposing an audit framework that is more responsive to current business needs. Using the traditional continuous auditing as a basis, the first essay proposes the predictive audit framework. The predictive audit is a forward looking process that utilizes predictive analytics to estimate possible outcomes of business activities, and allow auditors to execute their work proactively. The predictive audit differs from the traditional audit in several aspects such as control approach, objective, and frequency. The preventive audit is defined as a predictive audit with filtering rules to block highly probable faulty transactions prior to their execution. The second essay examines the application of the predictive audit on a bank’s real business data set to determine potential irregularities. This study aims to assist internal auditors concerning the validity of sales transactions. The possible outcome of the sale transaction is identified using three machine learning techniques: decision trees, logistic regression, and support vector machine. The results show that logistic regression outperforms other algorithms. With a proper sales variables selection, the predictive model could accurately predict results with high accuracy, true positive rates, as well as a reasonably low false positive rate. The robust results of the predictive audit can be used as a baseline to create screening rules for the preventive audit. In the third essay, the predictive audit is deployed to determine the possible results of credit card sales transactions. Consequently, the filtering rule constructs are derived from the predictive model. These rules can be implemented at the beginning of the business process as the preventive audit to flag or block transactions before they are executed. Alternatively, the filtering rules can be applied to the results of the predictive audit to reduce a number of transactions that auditors have to investigate. The rules significantly increase the possibility of discovering problematic transactions.