Data Analytics for Accounting & Finance 2021/2022

Please find below a set of sample questions from past exams.

Databases

Question 1

You have recently volunteered to help out with accounting and finance at your local    volleyball club. The club regularly receives match sales, as well as revenues from other sources, and it has been decided that some of these revenues can be returned to the  players of the club.

Each player in the volleyball team will therefore receive a small pay-out for playing. The pay- out will be dependent on the player’s contribution to a team, the leagues in which the team plays, and how many matches a team plays in every league.

You have been tasked to set up a database that carries enough information to allow the    calculation of these pay-outs. Another volunteer presents you with the following diagram, which depicts a data model that should help you with setting up the database.

You decide to implement your database in Excel. You note, however, that the model           contains two many-to-many relationships – it cannot be directly implemented into Excel in this way.

a)   Normalise the above model to remove the many-to-many relationships. Provide a brief commentary on the decisions you took during normalisation.

b)   Implement your normalised model in Excel. For each concept, mark the columns that are primary keys and foreign keys. Identify at least one extra column for every concept in      the model, and add at least three rows of imaginary data for every concept (e.g.,               imaginary player names, team names etc.).

Insert screenshots of each Excel worksheet in your document. Do not include the actual Excel spreadsheet file in the document. The screenshots are sufficient.

Please note that you do not have to calculate the player pay-outs in the Excel spreadsheet. The spreadsheet only needs to capture the data to calculate the pay-outs at a later time.

Add your candidate number to the Excel worksheets without further identifying yourself. Your candidate number must be visible in all screenshots.

c)   Another volunteer suggests that you use database software such as MS Access instead of spreadsheet software. Critically discuss the advantages and disadvantages of using   spreadsheet software such as Excel for the purposes of creating a database such as the one you have just created.

Question 2

You have started working at a local radio station, and have been provided with the following data model. The model is designed to help with accounting for royalties when artists have    their songs played on the radio.

a)   Normalise the model to remove the many-to-many relationships. Provide a brief commentary on the decisions you took during normalisation.

b)   Implement the normalised model in Excel. Identify a few relevant property names for each concept in the model, and add a few imaginary rows of data. Add              screenshots of your work. Include your candidate number in the screenshots       without further identifying yourself.

c)   Critically discuss the use of spreadsheet software such as Excel for the purposes of holding a database such as the one you have just created.

AI

Question 1

The following news item has appeared in the press.

LAST FALL, UNIVERSITY of Virginia computer science professor Vicente Ordóñez            noticed a pattern in some of the guesses made by image-recognition software he was building. “It would see a picture of a kitchen and more often than not associate it        with women, not men,” he says.

That got Ordóñez wondering whether he and other researchers were unconsciously injecting biases into their software. So he teamed up with colleagues to test two     large collections of labeled photos used to “train” image-recognition software.

Their results are illuminating. Two prominent research-image collections —including  one supported by Microsoft and Facebook —display a predictable gender bias in their

depiction of activities such as cooking and sports. Images of shopping and washing are linked to women, for example, while coaching and shooting are tied to men.

a.   The above news item gives an example of supervised AI. Explain this concept in your own words, using the algorithm above as an example.

b.    Explain how the gender bias occurs in the model, and how this bias can be detected.

c.    Critically discuss how you would address, and potentially resolve, any potential biases in an AI algorithm.

Question 2

University of XYZ (a fictional university) has developed an algorithm that predicts, for each student in the first year of university study, three types of student outcomes:

A. Drop out (student leaves the university)

B. Repeat-year (student must repeat the first year)

C. Pass (student passes the first year and continues to second year).

To predict these student outcomes, the algorithm uses student background data as well as student engagement data from the first three months of being at university. This includes  library activity, login data from the online learning environment, as well as public Twitter   activity.

The algorithm has been trained on a large set of student data, and to validate the success of the algorithm, the algorithm is run on a test set. The below table is an excerpt from the         result.

a.   Explain the table above in your own words. What is the usual name of this table, and what do the values of the cells mean? What does the note mean?

b.   Critically discuss the performance of the algorithm, and suggest ways to improve the performance.

c.   The University has proposed the introduction of a compulsory mentoring                     programme for those students who the algorithm has predicted will drop-out. Write a brief note below critically discussing this proposal and provide recommendations.

Enterprise Systems

Question 1

a)   Explain the steps involved into the procure-to-pay process. For each step, describe clearly which documents are exchanged between the companies involved and their departments.

b)   The procure-to-pay process can be supported in two ways: either by using separate systems for each department involved, or by using a single, integrated enterprise    system. Critically discuss the advantages and disadvantages of each scenario.

Python

Question 1

You work in the internal audit department of a company that has an online expense claim system. There are many expense claims every month, and the audit department can only check a small number due to staff shortages.

The department has recently encountered a number of incidents where employees were  fraudulently claiming non-existent expenses. These expense claims were typically entered online by employees with the free-text description ‘discretionary payment’ .

Your manager has been tasked to identify whether these incidents were one-off or part of a more widespread problem.

a) Explain the concept of “ population-based auditing” and discuss how this concept might help your manager with the above issue.

b) The audit department has decided that all online expense claims that contain the word ‘discretionary’ in the description should be marked as ‘suspicious’, and subject to further scrutiny. You suggest that a Python program may be helpful with this task.

Write down an algorithm that filters out the suspicious expense claims using Python              pseudo-code. Pseudo-code just refers to the broad structure or skeleton of the algorithm. It is fine if the details are missing.

You do not need to implement this Python program in a Python computing environment such as Colab or Jupyter.

c) Critically discuss at least two advantages and at least two disadvantages of using IT for these types of activities at the audit department.

Question 2

You have recently started working as an internal auditor for a company that sells dairy products.

You decide to focus on the procure-to-pay process, and you have asked the IT department  for an extract of transaction data covering purchase orders, goods received documents and invoices.