HR Analytics:

Batch 1
25-26, April 2022
 HR Analytics: The Relevance
List out three reasons that
explains the differences in
engagement levels of
Manager (Pers) & Dy.
Managers (Pers) of CIL
under last two directors
(in terms of importance)
List out three reasons
that explain the
differences raw coal
offtake during 2019-20
and 2020-21
(in terms of importance)
List out three reasons
that explain the variation
in profits of CIL across
2019-20 & 2020-21
(in terms of importance)

Ref: (Page 36, 41, 24-25)

https://www.coalindia.in/media/documents/Annual_Report_English_Deluxe_UiO2IIw.pdf
 Intuition Vs DDDM
• According to some studies, more
than half of Americans rely on
their “gut” in order to decide
what to believe, even when they
are confronted with evidence
that speaks to the contrary
• Data-driven decision-
making (sometimes abbreviated
as DDDM) is the process of using
data to inform your decision-
making process and validate a
course of action before
committing to it.
• Data is Omnipresent
• Analytics 1.0, 2.0, 3.0
https://blog.taskque.com/gut-feeling-helps-you-make-complex-business-decisions/, https://online.hbs.edu/blog/post/data-driven-decision-making
 Applications of Analytics
Reasons Behind Successful “Raids”

• Whether Amay Patnaik

conduct raids based on
anonymous tips alone?
• How is he resisting the
attempts to influence
‘from the top’

DATA Based Decision Making

than intuition/experience/gut feeling
 Applications of Analytics
How are we fighting against “CORONA”?

• Try to recollect different

types of data used in our
fight against Corona
– Daily Statistics
– Predicting the trends
• Who has the highest risk?
• How to determine the
containment zone
• Lockout decisions
• Developing vaccines
• Initial vaccination is for ___
• Gap between vaccinations

Using Big Data in Decision Making

https://www.youtube.com/watch?v=aorS66KhVOU https://www.youtube.com/watch?v=7liA2VWzCQA https://www.youtube.com/watch?v=05jwyl8b-_c
 BIG DATA in HR Domain

• There is No Such thing as Big Data in HR by Peter Cappelli (2017)

• https://hbr.org/2017/06/theres-no-such-thing-as-big-data-in-hr
– Does HR has access to relevant data?
– Average size of an organization
– Constraints in movement of HR data
Volume: run into petabytes (1015) or
Exabyte (1016 bytes) these days
AI in HRM: Challenges and Path
Forward (California Mgmt. Review 2019)
 Ease of Measurement

Market

Physical Operational

Human
Financial
Easier Difficult
 From What to measure
to How to Measure?

❖Problems associated with measuring “attitude”

❖Tips to measure intangibles:

➢ Decompose to components
➢ Secondary research
➢ Basic methods of observation and choosing
appropriate method to observe
➢ Measure just enough How to
measure
➢ Consider the error
the cost of
➢ Choose and design the instrument turnover?
Ref: “The transition: From what to measure to how to measure” by Douglas, 2010)
 Measuring the Culture

• Which one is easy to

measure?
− Culture
− Engagement

Internal
• Measuring Culture
through email records
to predict:
– Attrition
– Career growth Dominant Characteristics, Leadership Style
Mgmt. of Employees, Organizational Glue,
– Other usages
Strategic emphasis, Criteria of success
Ref: www.ocai-online.com
 People / HR Analytics
Staffing Rewards Accident
& & Engage- Revenue Fraud
Leadership
Workforce Investme ment Productivity Downtime
planning nts

Change
Perform-
Employee & other Innovation Managing
ance & Retention
Data Interven- Stakeholder
Mobility
tions

Travel, Organizational Network Analysis, Sentiments, Voice, Ethics, Admin

Data Management, IT, Business Consulting

Groundbreaking new insights & tools for better

evidence based decisions 10
Adapted Bersen Slide
 HR Analytics:
How HR may function in Future?*

Sometime in Future: You are still working; may be from

home or some remote sites. All administrative work will be
handled by computers. HR, if it still exists, focus mostly on
Human capital planning and organizational effectiveness.
Imagine for a moment that you are a newly hired recruiter and you are
attending a one-on-one termination meeting with your coach. She has
brought you in to learn by shadowing her. The employee who is being
terminated, Joe, enters the room.
He knows he is being let go but is reluctant to leave as of now. He asks
several questions about benefits and then leaves the meeting and the
organization. Although you are new to the organization, you know him.
He has a reputation as is a solid performer. He is friendly in a
professional way and is considered an asset to his team. When he
leaves, your coach asks, “Do you have any questions?”
“Just one,” you answer. “Why did you fire him? He is a good performer.”

Prof. Gloryson Chalil *Fitz-enz, Jac; Mattox, John (2014): Predictive Analytics for Human Resources
 HR Analytics:
How HR may function in Future?*

“Yes,” she says, “He is a good performer. But he is not a great performer.
He used to be great, and that’s why he has been here for seven years.”
As she talks, she types on her computer. She is opening a file for you to
review. “His stellar performance lasted through year five when he was
promoted and received a large salary increase. During the last two years,
he has plateaued as an average performer.
Another junior colleague got the promotion he was targeting and it further
reduced his performance in current job. We discussed this during last
talent audit at senior level and had counselling sessions with him.”
“So, average performance in a new role is a reason to fire someone?”
“No, but this file shows us that we have many options. He refused to
register for a partly paid competency development program, critical for
his job. As you know, we will not fully sponsor an employee for same
program in consecutive years. So I received an alert last month that Joe
has 70% chance to leave us and we discussed about him during last
talent audit, followed by a stay interview and counselling.

Prof. Gloryson Chalil *Fitz-enz, Jac; Mattox, John (2014): Predictive Analytics for Human Resources
 HR Analytics:
How HR may function in Future?*

Despite this, his interaction with the team and significant stakeholders has
reduced recently and therefore the probability became 78%.
While one of our competitor has opened a new position this month
beginning that matches Joe’s profile and we got a 85% warning last week.
Moreover It jumped to 91% when he requested a day off last Monday —
likely for a job interview. Again he asked for another leave yesterday, and
was insisting on availing it despite various pending critical works. He has
never done like this in his last seven years. As you know Mondays and
Fridays are usually interview days of our competitor and the probability has
jumped to 95%. If he had taken a day off on Wednesday, I would not have
been concerned this much...”
“You said there are options. Many options.”
“Yes.” She points to the screen. “I have an account with an online sourcing
system. I entered the job profile for Joe last week after getting 85% warning,
and there are 100 candidates that match the requisition.

Prof. Gloryson Chalil *Fitz-enz, Jac; Mattox, John (2014): Predictive Analytics for Human Resources
 HR Analytics:
How HR may function in Future?*

90% of them are in our metro area. Twenty of them make 25% less than
Joe, and have also 25% more relevant experience. Fifteen are men.
Five are women, and two of them are minorities. The colleagues on
Joe’s team indicated they would like to have a female leader. This is also
a chance to expand the diversity of the team.”
She points to a name on screen. “This is my top candidate. New hires
like to prove themselves, so she is likely to work harder than Joe, and
we expect that she will be 21% more productive. If we offer her the job
at a salary 19% lower than Joe’s, there is an 85% chance she will take
it.”
“But she lives outside the metro area. That will be long commute.”
“The probability jumps to 87% if we give her one flex day per week. It
jumps to 99%, if we let her work from home and only require her to
come to the office for critical meetings.”
Then you ask, “When do we interview her?”
“In 15 minutes.”

Prof. Gloryson Chalil *Fitz-enz, Jac; Mattox, John (2014): Predictive Analytics for Human Resources
 HR Analytics:
How HR may function in Future?*

“I better get ready then,” you say as you get up to leave the office.
“You’ll find her resume in your e-mail. Just sent it to you.”
Before you exit, you stop at the door and ask, “Can I see the report you
had about me prior to my job interview?”
She laughs. “Yes, I’ll share it with you during your performance review.
There’s a 90% chance you’ll stay with us long enough to have one.”

Reflection Questions (get ready to discuss during next class:

(1) Sub-functions of HR is included in the scenario we discussed right now
(2) Refer to the five components of HR Analytics. Whether the scenario
discussed here is perfect illustration of HR Analytics?
(3) Is the approach discussed is Ethical? … Philosophy of HR
(4) Where are we (practices in your firm)? What should be the target?
(5) How to customize the situation to your organization?

Prof. Gloryson Chalil *Fitz-enz, Jac; Mattox, John (2014): Predictive Analytics for Human Resources
 HR Analytics & HR Value Chain:
Evaluating the utility of common sense in a given context

Market Outcomes
Stock price

Financial Outcomes
Expenses, Profitability

Organizational Outcomes
Productivity, Quality

Employee Outcomes
Attitude, Behavior
 Aqua-guard Installation in Tata Motors

Identify Fixed, Variable and Opportunity Cost

of Aqua guard installation in Tata Motors
 Group Work
• Identify an HR initiative with positive ROI from CIL
• Identify an HR initiative with negative ROI from CIL
– Identify
• the fixed cost
• variable cost &
• opportunity cost of chosen initiative
– For further discussions:
• What are the possible reasons for negative ROI
• How will we choose investment opportunities with positive ROI
• Advantage of using ROI parameters in HR decisions/investments
 HR Analytics
An Illustration from the Past (1992-95)
Customer
Attitude Recomm-
Service/ endations
about Helpfulness
Job

Employee ROA
behavior Customer Op. Margin
Impression Rev. Growth
Attitude Merchandise
about /Value
company
Employee Customer
Retention Retention

5 UNIT 1.2 UNIT 0.5 UNIT

 Moneyball in Indian Context
Another Illustration from sports

Who will win

World T20
Super 12 Match
between India
and Pakistan
on Oct 23, 2022
at Melbourne?

– Group Exercise
• What data you should collect to predict the winner
– List out to share later
• Are you interested only in predicting the winner
or…(specify)?
 People/HR Analytics
Another Illustration from sports

– What should be the data for our analysis to predict

the winner of Oct 23, 2022 Match?
❖ All cricket matches between India and Pakistan or for a certain
period? What are the specific concerns?
❖ Which format?
❖ Which location?
❖ Do we have enough data?
❖ How to make sure that data is sufficient enough
❖ Are we interested only in predicting winner? If not what else?
❖ Usage of logic and contextual knowledge in data collection
 HR Analytics:
The Way to Proceed

HR Analytics more of a logical and contextual approach than a statistical approach

 Test your understanding
• Pakistan win 71% of matches while Afridi
Identify the concede less than 5 RPO
differences • To win the match Afridi should try to
between two contain runs than taking wickets
statements • In order to increase the sales turnover of
based on three the by 0.5%, Sears may consider investing
levels in HR in improving employee attitude by 5%
Analytic • Increase in employee attitude (from 80 to
Continuum 85), will increase customer attitude by 1.2%
in Sears

Does it mean:
• Any one bowler reporting good economy rate will ensure his team’s success?
• Increase in employee attitude will increase customer attitude in every firm?
 Data Capturing: Changes over a period of time

24
https://joshbersin.com/2020/08/is-the-employee-survey-dead-nope-its-becoming-smarter-by-the-minute/
 Diversity Analysis of CEO/COO
Positions in a Conglomerate
• You are assigned the responsibility of conducting a
diversity analysis of top executives (CEO/COO) in a
conglomerate. This conglomerate operates in three
continents and own ten industries of various sizes, across
different sectors.
• Organization has shared following data and you are
expected to make recommendations based on HR
analytics. Your recommendations should be based on
provided data and convincing enough to address any of
the issue identified by you.
– Descriptive analytics: Identify the trends & patterns related to
the background of (a) CEOs (b) COOs and inferences you
are able to make
 Diversity Analysis of CEO/COO
Positions in a Conglomerate
Highest No. of No. of
Sl Experience Educational Industry Functional Current Salary
No in Years Qualification Ethnicity Exposures Exposures Designation (Million $)
1 18 MBA Asian 3 4 CEO 12.1
2 13 M Tech European 4 2 COO 9
3 15 MBA African 2 4 COO 6.5
4 20 PhD Asian 4 4 CEO 7
5 13 Post Graduate European 4 2 COO 8
6 25 CFA Asian 3 5 CEO 9
7 17 MBA American 2 3 COO 10
8 25 PhD American 3 2 COO 9
9 18 Post Graduate African 3 3 COO 7.5
10 12 Post Graduate Asian 4 4 CEO 10
11 20 MBA European 4 4 CEO 10
12 16 MBA Asian 2 2 COO 8
13 21 M Tech Asian 3 4 CEO 10
14 19 MBA Asian 4 3 CEO 9
15 17 MBA European 2 3 COO 10
16 15 CFA American 4 5 CEO 9
17 18 Post Graduate Asian 5 1 COO 10
18 15 MBA American 4 1 COO 8
19 20 PhD Asian 3 4 CEO 13
20 13 MBA European 2 5 CEO 11
 Diversity Analysis of CEO/COO
Positions in a Conglomerate
• You are assigned the responsibility of conducting a diversity
analysis of top executives (CEO/COO) in a conglomerate. This
conglomerate operates in three continents and own ten industries
of various sizes, across different sectors.
• Organization has shared following data and you are expected to
make recommendations based on HR analytics. Your
recommendations should be based on provided data and
convincing enough to address any of the issue identified by you.
– Descriptive analytics:
– Predictive Analytics: By going through available data, make a
prediction about the minimum requirement in this organization to
become a CEO
• Hint: All CEOs will fulfil the minimum requirement, while none of the COO’s.
– Prescriptive Analytics: Make recommendations to improve diversity
level among CEOs: Short term, Medium term and Long term
 Class Exercise

Work with the shared data and try to cover three levels of
analytics:
Descriptive analytics:
Predictive analytics:
Prescriptive analytics:
 Class Exercise

Avg Sp_com Tech Bus

No Pr_trng Pst_trng p Comp Comp Eng_90 Eng_1 yr Bill_rate
LP
MP
HP

Avg
Pre_tr Tech
No ng Post_trng Sp_comp Comp Bus Comp Eng_90 Eng_1 yr Bill_rate
LP 12(4) 20.17 27.17 94.5 50% 50% 0.443333 43.00 20%
MP 74(14) 30.51 41.62 82.83 85% 90% 88% 88.00 41%
HP 17 (11) 42.29 46.14 27.44 95% 96% 95% 66.00 83%

Based on HCMI framework (https://www.hcminst.com)

 Think Like a Data Scientist
Class Exercise

• How serious are late

starting of meetings in your
organization?
– How will you go through
descriptive, predictive and
prescriptive analytics based
on your learning from the Reflect about meeting you were a part of in
course so far… the organization and define the following:
– Remember we are yet to – Define actual starting of the meeting
develop complex metrics,
– How much of data is required?
study analytic tools and
optimization techniques – How will you analyze the data that
will help you to:
• Make predications
• Make prescriptions
 HR Analytic Process *
• Step 1: Frame the Central Problem
– Interview Key stakeholders
– Review existing documents that provide the context
Compare: Late commencement of the meetings (Theme A) and meeting
that are not getting finished on time (Theme B). How are they different?

• Step 2: Apply the conceptual model to guide the

analysis
– Identify HR/people and business variables that are
likely to have association with the problem outcome
– Be alert to idiosyncratic events and additional data
that could be relevant
Exercise: Differences in conceptual models of two themes?
*Hoffmann et al.(2012): Calculating Success
 Types of Variables
• Control variable: a factor in an experiment which must
be held constant.
• Dependent variable: the outcome of an experiment. As
you change the independent variable, you watch what
happens to the dependent variable.
• Independent variable: a variable that is not affected by
anything that you, the researcher, does. Usually plotted
on the x-axis.
• Mediating variable: variables that explain how the
relationship between variables happens.
• Moderating variable: changes the strength of an effect
between independent and dependent variables.

*Hoffmann et al.(2012): Calculating Success

 Exercise (Individual)

Categorize following variables to DV, IV & CV (Mediating and

moderating variables, if applicable):
➢ Acceptance of Offer, Job Offer, Joined or Not, Terms of offer

• What are the possible control

variables for this study
Terms of • Can you identify other
Offer moderating variables

Job Acceptance Joined

Offer of Offer or Not

• Identify an example with multiple independent variables

• Is it possible to have multiple dependent variables
 CENTRAL PROBLEM
• Lack of Knowledge Management Infrastructure (KMI)
leads to reduced Employee Performance in case of
employees going through a transition
DEPENDENT VARIABLE Mediator VARIABLE
• Employee Performance • Willingness to use KMI
INDEPENDENT VARIABLE CONTROL VARIABLES
• Quality of Knowledge • Place of posting
Management Infrastructure • Duration at the place
• Level
MODERATOR VARIABLE • Exposure to similar assignments
• Transition (Yes/No) • Demographic controls
Customise this topic to Union management ability of Personnel managers through KMI
Not necessary to have moderator/mediators in every study
 HR Analytic Process *
• Step 3: Capture relevant data
– Pursue relevant data across all departments, who can
grant you access to relevant data
– Reconcile the differences in definitions, codes and
time frames
– Store valid data in Analytical database
• Step 4: Apply analytical models
– Employ appropriate formal quantitative techniques,
looking for suitable patterns over time
– Examine results and identify relevant models

*Hoffmann et al.(2012): Calculating Success: How the New Workplace Analytics Will Revitalize Your Organization
 HR Analytic Process *
• Step 5: Present statistical findings to stakeholders
– Managers from non-statistical background should be
able to appreciate the findings
– Validate and enrich patterns through follow up
interviews/focus group discussions
– Identify new problems and need for further
analysis
• Step 6: Define action steps to implement solutions
– Operationalize changes in policies, procedures that
result changes in workforce behaviour
– Monitor and document changes and outcomes

*Hoffmann et al.(2012): Calculating Success

 HR Analytic Process
From Concepts to Application

I. Frame the Central Problem

II. Apply the conceptual • Categorize these
steps into three
model to guide the analysis types of analytics:
III. Capture relevant data – Descriptive
IV. Apply analytical models – Predictive
V. Present statistical findings – Prescriptive
to stakeholders • Reflect on late
commencement of
VI. Define action steps to meeting scenario
implement solutions
 Ingredients of a Successful People
Analytic Project *
Ingredients Include Process such as
30% Gather clean and fresh data you can access, Partner with
Data finance, marketing, operational dept for data, external as
required
10 % Collect key hypotheses from relevant executives. Your
Stakeholdering analytic project should answer their business concerns
15% Using advanced statistical tools to develop insights. It can be
Analysis qualitative too
15% Create interest for management in your project and its
Story telling finding. Objective is to influence them and others involved in
HR Analytic project
20% Converting the insights into actions for the benefit of
Implementation organization. Use learning in future projects
10% Defining accountabilities, embedding reporting relationships
Embedment and timelines. Celebrate short term wins and be flexible, if
required.

*Fitz-enz, Jac; Mattox, John (2014): Predictive Analytics for Human Resources
 Monetary Value of Variations in Job
Performance (Judgment Based Approaches)
• The Cascio-Ramos Estimate of Performance in Dollars (CREPID)
For an IT Professional working in United States
Principal Time/ Impor- Relative Points Net Dollar
Total Dollar Value
Activity Frequency tance Weight Assigned Value
1 4 4 16 10.36 5178 1.35 6990
2 5 7 35 22.65 11327 1 11327
3 3 5 15 9.71 4854 1.25 6068
4 0.5 4 2 1.29 647 2 1294
5 2 7 14 9.06 4531 1 4531
6 5 5.5 27.5 17.80 8900 0.5 4450
7 6 4.5 27 17.48 8738 0.4 3495
8 3 6 18 11.65 5825 1.5 8738
154.5 100 50000 46893

• Identify three levels in HR Analytics during CREPID exercise?

• Minimum number of employees required for this analysis
• Not useful in situations wherein pay systems are mostly based on rank, tenure
or hourly pay rates (Like CIL) to determine the salary, but many more useful
applications.