Abstract

The internal audits of a company are a fundamental process for quality assurance and proper compliance with different processes within a company. For this reason, it is also important that companies are able to adequately plan their implementation; But it often happens that companies do not have the necessary tools to make these adequate plans. Within the topic of planning, in this context, the factors of effort of internal auditors in work hours and capacity (number of audits) are taken into account. These aspects are vital for the organization of times and audits over a period of time, since, if an adequate calculation is not made, it can significantly affect planning and trigger different consequences for the company and its processes.

Given the above, it was proposed to develop and implement predictive and prescriptive analytics models to facilitate the planning and optimization of internal audits in terms of hours and number of audits (2 models); improving an (existing) digital planning tool.

Historical audit information was obtained, worked and processed within an IT company. Different Machine Learning Python libraries were applied to carry out the modeling process, which include: PyCaret, H20, FLAML, TPOT, LazyRegressor, among others.

Various evaluation metrics such as coefficient of determination (R2), root mean square error (RMSE), mean square error (MSE), and mean absolute percentage error (MAPE) were used to evaluate the models. The best results were obtained by Random Forest models, where R2 metrics of 94.40% and a MAPE of 6.26% were obtained for the hours of effort model; An R2 of 90.49% and MAPE of 6.77% were obtained for the capacity model.