ACST3032
Actuarial Data Analysis
Semester 2, 2023
Assignment Two
This assignment is worth 20% of your final grade.
this assignment is out of 55 marks.
Background
You are an actuary working in the analytics team at TrueSight, a management consulting firm. Recently,
your firm has been approached by the Human Resources (HR) team of Providence Bank, a medium-sized
bank in Ozland.
Providence Bank has been experiencing relatively high attrition rates in its Financial Analytics Division (FAD).
Frequent employee turnover significantly impacts the bank's morale, and the cost of hiring new employees
often exceeds that of retaining existing ones. The HR team seeks to understand the factors that encourage
employees to stay with the company and those that prompt others to leave. This understanding is
imperative to prevent further loss of valuable employees.
Question
Your correspondence with Providence Bank is through William, the HR retention manager of Providence
Bank. William has provided you with de-identified datasets containing the HR records of employees in the
Financial Analytics Division (emp-record.csv) and the survey results of the employee (emp-survey.csv).1
William commented that, based on his general experience, the top 5 reasons for employee attrition are:
1. Inadequate compensation and benefits;
2. Poor work-life balance;
3. Poor leadership and management;
4. Lack of career growth; and
5. Poor workplace culture.
After several meetings with William, you confirmed the specifics of the business deliverables for this project.
You need to conduct an exploratory analysis and investigate the performance of various models before
arriving at a final recommended model. Additionally, you need to provide a counterfactual analysis to
estimate the number of employee attritions in the FAD this year under several retention strategies (please
refer to the report outline for more details). You also noticed that William has limited technical proficiency
in statistics.
The proposed outline for the report is below. The report should not exceed 12 pages excluding the appendix
and the reference list.
1. Exploratory Analysis (15 marks)
This section should provide an overview of employee attrition in FAD and compare it to William’s general
experience. You should provide reasoning for your analysis steps and choices whenever appropriate.
Rubrics:
- 3 marks. An overview of the employee attrition in FAD
- 12 marks. Employee attrition analysis. Hint: you may find William’s comment useful here.
1 Providence Bank conducts an annual employee engagement survey via WorkerAmp, a HR analytics firm.
2. Predictive Models and Analysis (20 marks)
This section should include:
- Creation/engineering of appropriate response and predictor variables.
- Building and examination of two predictive models:
• A logistic regression or a penalised logistic regression; and
• A random forest model or a bagged tree model.2
- Explanation and justification of the choices of models. For example, if you wish to use a logistic
regression rather than a penalized logistic regression or vice versa, you should justify your choice.
- Explanation and justification of the modelling steps and choices on the modelling components, such as
pre-processing steps (i.e., recipes), resampling techniques, performance metrics, hyperparameters, etc.
- Discussion of the strengths and the weaknesses of the two chosen models for this exercise and
recommend a final model.
Rubrics:
- 15 marks. Descriptions and justification of modelling choices of models
- 5 marks. Recommendation (and its justification) of the final model
3. Counterfactual Retention Analysis (10 marks for ACST3032 students, 15 marks for ACST4062/6032
students)
This section should include a counterfactual retention analysis requested by William.3 He requested a
counterfactual retention analysis for two retention strategies:
- Competitive pay. Increase the compensation for all employees by + 2% where represents the
percentage increase of compensation from 2021 to 2022.
- Reduced overtime. Limit overtime so that the average monthly overtime does not exceed 5 hours
for each employee in 2022.
Using the final predictive model in Section 2, estimate the counterfactual number of employee attritions in
FAD at the beginning of 2023 for three scenarios: without any retention strategy (base case) and under the
two retention strategies mentioned above. You should state any relevant assumption used in the estimation.
In addition to the above two strategies, propose an additional retention strategy and estimate the
number of employee attrition under this new strategy. You should also provide reasoning for your
proposal.
Rubrics:
- 10 marks. Estimated number of employee attritions under each of the three scenarios (base case,
competitive pay and reduced overtime).
- 5 marks. Proposal and estimation of the third retention strategy.
2 Depending on the computing power of your computer, model tuning can be very time-consuming for ensemble
models. To reduce the number of hyperparameters to be tuned, you may use `trees = 500` and assume that there is
no noticeable improvement in performance for `trees > 500`.
3 A counterfactual analysis explores outcomes that did not actually occur, but which could have occurred under
different scenarios.
Overall Presentation and communication (5 marks)
The content of the report contributes a total of 45 marks (or 50 marks for ACST4062 and ACST6032 students).
The report presentation and communication contribute 5 marks.
Appendix A - Data Description
Variable Description
id The anonymised ID number of the employees.
age The age of the employees.
gender The gender of the employees.
education The education level of the employees.
location The office location of the employees (the headquarter of Providence Bank is in
Sydney).
distance_from_home The normalised distance between the employee’s home address and the office
location.
remote Whether the employee works remotely.
compensation The compensation of the employees in 2022.
percent_hike The percentage increment of compensation for the employee from 2021 to 2022.
level The professional level of the employees in FAD, which comprises of Analysts and
Specialist (specific sub-level information is excluded for privacy and identification
reasons).
tenure The length of employment in years at the current position.
promotion_last_year Whether the employee was promoted internally (including sub-level promotion)
within the last year.
monthly_overtime_hrs The average monthly overtime hours of employees in 2022.
status The employment status at the beginning of 2023: Active or Inactive (i.e. left the
job).
Table 1: Variables for the dataset `emp-record.csv `.
Variable Description
id The anonymised ID number of the employees.
theme The key indices of the annual employee satisfaction survey are provided by
WorkerAmp, a third-party HR analytics company. These indices reflect employee’s
satisfaction in the following aspects of their job:
• career – the career growth and professional development,
• work – their work life balance, and
• leader – their attitude towards current leadership and management.
rating 1 – Very unsatisfactory
2 – Unsatisfactory
3 – Neither satisfactory nor unsatisfactory
4 – Satisfactory
5 – Very Satisfactory
Table 2: Variables for the dataset `emp-survey.csv `.
End of question
Further instructions:
Cover page: Please include a cover sheet in your submission. The cover sheet is available on Wattle and it
does not count towards the page limit.
Page limit: Please submit your assignment in a word or PDF document not more than 12 pages. You should
think about how to best display your answers, workings and assumptions within this limit and marks will be
awarded for presentation and communication. You may (and should!) include relevant tables and graphs in
your document. Please ensure you explain your analysis and justify any modelling choices.
Appendix: Please submit the relevant R code used for modelling in the appendix. The appendix does not
count towards the page limit. The appendix will not be marked but might be checked to clarify a response
in the report if necessary. There is no need to include other contents in the Appendix section.
Reference: You do not need to reference any other material to complete this assignment but if you do,
please ensure you properly reference your work. You must adhere to appropriate practices regarding
referencing the work of others in any work that you do. Accepted academic practice for referencing sources
that you use in presentations and assignments can be found via the links on the Wattle site.
AI tools: The use of ChatGPT to assist any answers in this assignment is not allowed. Any detected use of
ChatGPT or other similar artificial intelligence tools will be referred to the University Registrar as a breach
of academic integrity.
End of assignment
Note: All entities depicted in this assignment are entirely fictitious.