Data Science for Auditing

The course unit articulates interdisciplinary concepts of mathematics and statistics, applied to quantitative modelling.

The classes are, in accordance with the curriculum, theoretical-practical, planned and prepared to actively involve students at various moments or throughout the entire class.

In the theoretical part, which introduces concepts, fundamental results and methods, the expository method will tend to be used, interspersed with tasks that encourage more active participation from all students. These tasks include asking questions to and by students, either orally and/or on a platform, and also proposing debate/discussion in small groups on some aspect/topic presented.

The practical part will be devoted to the full development of the listed skills, through the commented exemplification of procedures and/or problem solving under the guidance/tutoring of the teacher, encouraging independent work or work in small groups. There will be a strong interaction between theory and practice, giving, whenever possible, a central role to the visualisation and treatment of concrete and real situations.

Good monitoring of classes by students requires regular attendance and a willingness to remain involved beyond the classroom, with the start or completion of tasks agreed in class.

The increasing availability of data and the incorporation of analytical techniques into audit processes require the mastery of quantitative methods that enable the analysis of patterns, the identification of deviations, and the support of the formulation of quantitative expectations. In this course unit, the focus is on the application of data science methods to auditing, with particular emphasis on forecasting using time series and regression models, as instruments supporting audit procedures.

Intended learning outcomes:

Upon successful completion of this course unit, students will be able to:

• Identify and contextualise opportunities for the use of quantitative forecasting methods in an auditing context.
• Select and apply time series and regression methods for the formulation of auditable quantitative expectations.
• Assess, based on statistical criteria, the adequacy and limitations of the predictive models used.

Skills to developed:

• Identify datasets and quantitative problems relevant to the application of forecasting methods in a auditing context.
• Prepare and conduct exploratory data analysis, whether using cross-sectional data or time-series data.
• Build and estimate time series and regression models with a view to generating quantitative expectations.
• Validate the statistical assumptions of the models used and assess their adequacy and limitations.
• Evaluate the predictive performance of models based on appropriate statistical criteria.
• Use software tools to support data analysis and quantitative forecasting.

1. Introduction to forecasting methods
    1.1. General considerations about forecasting methods
    1.2. Qualitative methods vs quantitative methods
    1.3. Causal and non-causal quantitative methods
    1.4. Exploratory analysis data and instruments
    1.5. Basic steps, concepts and tools in quantitative forecasting

2. Model-based forecasting for time series
    2.1. Forecasting with elementary methods
    2.2. Forecasting with exponential smoothing methods

3. Forecast based on econometric models
    3.1. Linear regression model: specification, extensions and assumptions
    3.2. Estimation by the method of least squares
    3.3. Model evaluation
    3.4. Selection of predictors
    3.5. Forecast: forecast of the isolated value and the average value; forecast ranges

NAO