Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
BEM2031 Assessment
Overview of Summative Assessment :video of which will help you in
guiding and having an overview of what the assessment is all about.
You can download the le here and read the assessment brief.
There are two components to consider 1) the outline for the critique
is expected two weeks before the full report but may be sent to me
earlier, and 2) the full critique of the report along with your
additional analyses. There will be an assignment with the data, some
basic scripts, and all the packages loaded.
HR_Analytics
Archive le with a PDF of the report and the
data.
kaggle_hr_analytics.csv
A CSV le with the data for the reports.
Description of the columns can be found
below.
don-t-know-why-
employees-leave-read-
this.pdf
don-t-know-why-
employees-leave-read-
this.html
This is a personal reproduction of the
Kaggle Kernel found here. This report was a
“Report of the Week” in 2017. The creator
was awarded $500 for this distinction. But
there are many aws in this analysis.
Use the attached PDF or HTML les, not the
online version. The online version is missing
some things. The PDF doesn’t have the
interactive chart, so the HTML is
recommended.
Outline for Critique
Length: 300-500 words
Weight: formative
Deadline: 11-03 -2022 15:00:00 GMT
Feedback: In a group setting( maybe Teams)
Due two weeks prior to the deadline for the report critique. I am
looking to see that you have thought through the assignment and
what you plan to write, analyze, or create in response. This can be
delivered as an outline with bullet points, or with complete thoughts
in well-reasoned paragraphs.
Analytics Report Critique
Word Count: 3,000 words
This assessment is divided into three primary tasks. See the marking
table at the end for more details.
1. You must critique the given report. See the points of critique
below.
2. You must provide at least one new predictive model of your
own. Assess the t of the model, justify your approach, and
interpret the ndings.
3. You must create at least one new visualization of your own.
You must justify your visualization approach, and provide an
interpretation of what you learn from it.
Context:
Consider you lead an analytics team at an international corporation.
There have been a number of notable departures of your top talent
recently - some of the best contributors have quit their jobs and left
for new companies. The top management is concerned that
something is systematically wrong with their retention policies. You
have been collecting data on job satisfaction, performance, and
other metrics for years and merged this information with other data
on employee work loads and other information in order to
determine what is leading to employees to leave the company.
The report found in don-t-know-why-employees-leave-read-this.html
is the report that was delivered to you by a member of your data
science team. It is heavy on analytics and visualization, but very light
on interpretation and explanation since it was intended to be a
discussion piece for a series of meetings that will happen soon. You
are to read the report and come to the meeting prepared to ask a
number of questions about the analysis, ask for changes, and show
some of your own results for comparison.
The data science team was able to produce a dataset,
kaggle_hr_analytics.csv, consisting of the following features from
15,000 employees.
Employee satisfaction level, based on survey data
(satisfaction_level)
Last evaluation, supervisor rated performance evaluation
(last_evaluation)
Number of projects employee worked on (number_project)
Average monthly hours (average_montly_hours)
Time spent at the company in number of years
(time_spend_company)
Whether they have had a work accident (1 = yes, 0 = no)
Whether they have had a promotion in the last 5 years (1 = yes, 0
= no)
Department, text data based on the dierent departments
Salary, are they highly paid, medium paid, or low paid.
Whether the employee has left (1 = yes, 0 = no)
How to get started
1. Read the report, try and understand it as much as you can.
Take notes. Determine where the report is probably lacking or
potentially made some mistakes. Ask me questions when you
get a chance.
2. Play with the data. Look at the data and try to lter it, create
some of the charts, maybe. I don't know.
3. Read the assignment brief, especially the list of questions I
prompt you to think about. Consider the business problem -
why are people quitting their job? And how well does the data,
analysis, etc. address this problem?
Your Predictive Model and Visualization
You have a lot of freedom regarding your choice of visualization or
predictive model. You can use code directly from the code sharing
document. Whatever choice you make you should do three things to
contextualize and explain your decisions and results:
1. Say what you're going to and why decided to do it. What
motivated your choice of visual or model?
2. Say what you did, and how. What challenges did you face and
how did you solve them? Challenges could be conceptual or
technical.
3. Say what you learned and what value it brings to the analysis.
How do the results alter your perspective on the original
business problem?
Additional Details
This data was part of a Kaggle competition to predict who would
leave and who would stay based on the data. The original Kaggle
report by Yassine Ghouzam can be found here. You are encouraged
to explore the comments on Kaggle or other reports and analytics
that have used this dataset, although the original dataset has been
removed from Kaggle.
Refer to the Proposal Review Guide in Appendix A in your Data
Science for Business book for a good outline of what to look for and
what to critique in the report. There is another sample report and
critique in Appendix B as well.
The document should be prepared in word processing program (e.g.
Microsoft Word or Google Docs) and submitted as a PDF or DOCX
format. Copy and paste all tables and visualizations into the
document. You are not required to include the code you wrote to
produce the analyses or visualizations. If you do choose to include
the code, add it to the end of the document in an appendix. The
visualizations, tables, and code do not count towards the word
count.
You do not have a required quota for academic references. They are
not expected to add academic references to your report. Support
your claims using a sound analysis of the data provided. If you do
reference material outside the reports or data described here, you
are expected to cite that material. Use APA style for your references
if you include them.
A good way to structure the report is to consider the CRISP-DM - the
CRoss Industry Standard Process for Data Mining.
Structuring Your Report
Your goal is to be critical of the report that was given to you by
another team member. You need to consider ways in which this
analysis could be improved and provide your own interpretation of
the situation before this report is shared with company leadership.
The questions listed in each heading are suggestions. You do not
need to answer each one, and your report can explore other
questions not listed here.
Business Understanding
What is your understanding of the goals of this project? Is the
data analysis suitable and the data used going to be to be to
help guide decision making? What are the costs and benets
of this analysis? Who do you think will be harmed or beneted
by this analysis?
Data Understanding.
Is the data appropriate? What don’t we know from the data
that would be helpful when understanding the results? What
data should they have included that was missing?
How eective are the visualizations at building the narrative of
the report? How could they be improved? What visualizations
are missing that could help?
Data Preparation
Did the report appropriately explore all the dierent ways in
which the data may be corrupted? What were additional
cleaning steps they could have considered? Should they have
reshaped the data in any way? Do you trust the data? What
would make you trust or distrust the data?
Modeling
Were the analytics choices here appropriate? Did they apply
them correctly? You don’t need to know the specics of the
code, but more about the general approach (e.g. was a
decision tree a good choice, or is there another analysis that
would have been better used?)
How were the models evaluated? How do you know that they
t the data appropriately? What approaches did they use to
avoid overtting? Do we know if these models will work on
unknown data in the future? What metrics could they have
used to assess the quality of the model?
What were the important variables in the models and how do
you know they are important? Do we know how these variable
impact the outcome? How could they have measured that
impact?
Evaluation
Should this project move on to deployment? Do you think that
the process here, the data used, the analytics and
visualizations produced helped solve the problem? What
Module information Access Your Library
changes to the process would you make to continue this
assessment in the future? What was missing from the data or
analytic process that you feel should be included?
Deployment
What action would you recommend considering the results of
the report and your own analyses? What makes you think
these actions would be feasible and eective? Note - you don’t
have to be an expert in human resource management (HRM),
you don’t have to justify your decisions based on a deep
understanding of HRM, you should justify it based on what the
data and analytics suggest. Even if you do have a good idea of
HRM, I expect you to justify any actions using analytics based
on this data.
How would you deploy this analysis in an ongoing process?
What are the needs of a descriptive, predictive, or prescriptive
deployment? What would you need to do to implement an
ongoing prescriptive dashboard system? What consequences
do you expect from these analyses?