This Faculty Spotlight is on Dr. Nuriddin Tojiboyev, Assistant Professor, Accounting and Information Systems. Nuriddin joined TCNJ School of Business in the fall of 2022. His research interests include developing data analytics models that contribute to the effectiveness and efficiency of auditing. Currently, he is researching the topic, “Rule Induction from Exceptions and Outliers for Continuous Audit.” We recently spoke to him about his research, and this is what he had to say.
Can you tell us a little about this topic?
This research proposes applying semi-supervised rule induction algorithms to the confirmed instances of misstatements detected by investigations of exceptions and outliers in Continuous Audit. The proposed modification to the framework of Continuous Audit allows updating of existing rules or generating new rules for additional filter discovery. This addition to the analytics of the continuous audit advises auditors about risks they are unaware of and recommends risk metrics that should be added to the transaction verification module. The proposed artifact improves the utilization of risk metrics in transaction verification, allowing the users to decide how deeply they prefer to focus on balanced or unbalanced risk metric values. The analyses of confirmed instances of outliers inform the users about additional useful rules to catch exceptions, the rules that the users are not aware of.
What drew you to choose this topic?
The application of data analytics models to existing audit models and frameworks is an interesting topic for design science researchers in this domain. However, the rule induction from identified and investigated exceptions and outliers is not a well-researched area in audit analytics. Such research would promote the application of rule induction in other audit analytics models.
Can you describe any challenges you have found in conducting this research?
The simulations for the evaluation of the designed artifact require extensions of the rule induction classifiers in the programming languages. However, there is only one rule induction classifier (DecisionTreeClassifier) available in sklearn (Scikit-Learn) machine learning library software. Furthermore, this classifier follows the “separate and cover” strategy rather than the strategy “divide and conquer”, which would have been ideal for this research. These challenges require extra coding from researchers who use rule induction in their data analytics models.
What is the most interesting fact about this area of research?
The limited audit budget allows investigations to be carried out on only a small part of the population. However, continuous audit requires transaction verification applied on a full population of items. This requires a data analytics method to induce rules from a partially labeled dataset. Although rarely used in audit analytics, semi-supervised learning allows researchers to use classifier algorithms on partially labeled data. Using rule induction algorithms in semi-supervised learning may ultimately provide a solution for limited audit budget constraints.
What direction do you think your research will take you in?
Properly induced rules from audit datasets might be able to evaluate the design and operating effectiveness of internal controls. I believe this research will contribute to the automated evaluation and modification of internal controls in audit.
Read more about Dr. Nuriddin Tojiboyev and what he is currently working on. Look for the next Faculty Spotlight soon!