HR Analyst Program

Vishak Bharadwaj P
IIT-IIM
IIM Alumnus, Strategic HR Partner, Monocept Consulting
HR Analyst Program: Agenda
ession Contents Plan
1 Building understanding of descriptive, predictive and prescriptive analytics for HR.
2 Promoting appreciation of the application of HR analytics to key HRM systems.
3 Providing overview of the statistical methods for predictive HR analytics.
4 Developing insights into analysing data to effectively manage talent in organisations.
5 Identifying areas in the HR life cycle where advanced analytical techniques can be
leveraged.
6 Learning the application of various analytical techniques across Recruitment,
Engagement and Retention.
7 Learning about the challenges of identifying appropriate measures/ HR metrics.
8 Applying the principles learnt to solve real-world
real cases.
hat is your expectation from today’s workshop?
Introduction to Peopl
Analytics-Why?
Why? – People Analytics Business and Decision-Making:
Decision
nal Reasons
etter Pay, Career path 1. Ability to make sense of more workforce data getting captu

etter Power 2. Building a business case using Cost Benefit Analysis

p/Society Reasons 3. Volume of data that needs to be

o standout among/conform with peers 4. Describing the situation

rend Reasons? 5. Diagnosing the problem
ncreased volume of data 6. Predicting the Future
7. Prescribing the best course of action
ources of
ower
 Building a business case using Cost Benefit Analysis
e Decision Issue, Relevant Stakeholders, Target Group and Objectives.
e issue or strategic problem. Issue must be
rategic or operational
the line of strategy or impacting strategic goals.
ate in the form of a question: "What is the best method for enhancing equipment maintenance skills among our operator work force
e Decision Options
valuating Various Vendors
uild vs Buy Decision
pportunity vs sunken cost: Total effort and resources that have been invested in a project to date
xed vs Variable cost
a) Fixed costs are economically relevant only if we have to spend money directly on them or
if they incur an opportunity cost elsewhere!
b) Amount of effort and resources required to complete the project
c) Present value of the cash flows that the project will generate
mpact on firms future income
 Building a business case using Cost Benefit Analysis
mining RoI by
Identify potential costs
Identify potential benefits
Compute the total payoff formula for each option
Calculate the ROI using an appropriate index
al Budgeting Process, Expenses and Capitalization.
Compute Breakeven,
Compute the decision rules for choosing among options
1
A day in the Chief People Officer of XYZ Ltd.
Building understanding of descriptive, Diagnostic, predictive
and prescriptive people analytics.
ing understanding of descriptive, Diagnostic, predictive and
riptive people analytics.
Descriptive Analytics
interpretation of historical data to better understand changes that have occurred in a business.
tive analytics describes the use of a range of historic data to draw comparisons. Most
nly reported financial metrics are a product of descriptive analytics.

es:
adcount report of all employees within the organization is a form of descriptive analytics.

1: Find the Descriptive statistics for the given data.

Diagnostic Analytics
stic analytics takes descriptive data a step further and provides deeper analysis to answer the
n: Why did this happen? Often, diagnostic analysis is referred to as root cause analysis. This
s using processes such as data discovery, data mining, and drill down and drill through.

es:
y Attrition is high in Junior Programmers?
Generation Z needs not met
Market pay is higher
They lack stability
Work is not interesting
They lack connect with the organization
They do not find a career path in the organization

2: Find the Diagnostic Analytics for the given data.

Predictive Analytics
ve analytics focus on what might happen in the future, based on the details of past events. This may
of which salespeople are likely to resign within the next 90 days. Predictive data is gained through
g, machine learning and artificial intelligence.
now what’s going to happen, you can prepare for it in advance. Much like if you know that it’s goin
w, you can pack an umbrella in your brief case tonight.

es:
o are the Junior programmers going to leave the organization in the next 90 days.
ich team will face the highest challenge in the next couple of years.

2: Lets Predict something from the given data.

Prescriptive Analytics
tive analytics suggest decision options and actions you can (should) take, based on the predictions.
urely human decisions that are often subject to illogical biases, the decisions recommended by presc
s are based on data, and therefore, more reliable.

es:
at steps needs to be taken to ensure attrition in Junior Programming
Define the new trainings for address upskilling issue
Address gaps with the market pay
Build a career path for Junior developers.

3: Do the Prescriptive analytics for the given data.

 Providing overview of
the statistical methods for
predictive HR analytics.
Fundamental of Statistic test
Test Statistic
Significance Level: Alpha, the significance level, is the probability that you
will make the mistake of rejecting the null hypothesis when in fact it is true.
P Value: The p-value
value measures the probability of getting a more extreme
value than the one you got from the experiment.
If the p-value
value is greater than alpha, you accept the null hypothesis. If it is
less than alpha, you reject the null hypothesis.
Upper value, Lower Value
Inference
Null and Alternate Hypothesis
The null hypothesis is a general statement that states that there is no relationship between two
phenomenons under consideration or that there is no association between two groups.
An alternative hypothesis is a statement that describes that there is a relationship between two
selected variables in a study.
When to reject null hypothesis?
When the p-value
value of the data is less than the significant level of the test, the null hypothesis is
rejected, indicating the test results are significant.
However, if the p-value
value is higher than the significant value, the null hypothesis is not rejected, and
the results are considered not significant.
Some sample null Hypothesis
Gender is not related to Performance
Performance is not related to Pay
Performance is not related to Engagement
Pay is not depended on Gender.
Causation and
• In statistics, the phrase "correlation does not imply causation"
Correlation refers to the inability to legitimately deduce a cause-and-effect
cause
relationship between two variables solely on the basis of an
observed association or correlation between them.
ons to ask for while choosing
stical Method
w many variables do we
e?
at’s the
ective(Descriptive/Classify/
mpare/Predict/Explain)
ch scale to measure
m?(Continous/Discrete/Ratio)
pendent/Independent
the sample autocorrelated
ocation/time?

statswithcats.files.wordpress.com/
Which test needs to be used when?

https://statswithcats.files.wordpress.com/
oosing
best
y to
ually
resent

alyticsvidhya.com
Learning about the challenges of identifying appropriate
measures/ HR metrics/Scorecards

Types of Metrics
• Leading indicators are much more likely to be unique to your
company, which makes them harder to build, measure and
benchmark.
• Lagging indicators tell you about what has already happened, with
common examples being revenue, profit and revenue growth.
Metrics and Dashboarding
mographic Indicators
Number of 4. Talent Acquisition
employees/Headcount a) Time to fill
Age, Gender, Location b) Cost to fill
Diversity ratios c) Quality of hire
d) Candidate Experience
ganisation Culture and e) Positions open/Positions
ucture Filled
Manager:Employee f) Offer Acceptance Rate
HR:Manager:Employee
% Permanent
g) New Hire turnover

oductivity 5. Talent Management

Absence Rate a) Internal Mobility Rate
Overtime Rate b) Promotion Rate
Output/Throughput per
person
HR Cost/Person
Profit/Employee
Metrics and Dashboarding
ng and Development 8. Career and Succession Management
Cost a) Bench Strength
a) RoDI b) High Potential Turnover
b) Total training cost c) Career Path Ratio(Promotion/Transfers)
Quality d) Career Plateau ratio
a) Trainer Feedback
b) Change in Behavior 9. Total Rewards
c) Abandonment rate a) Total Reward ratio(Monetary, Non
d) Training success rate monetary, Individual, Group)
Time b) Compa-Ratio(actual
Ratio(actual Salary/Midpoint
a) Number of hours of Salary)
training hours provided 10. Engagement
b) Average training a) Engagement Index
duration b) Turnover/Attrition Rate
Volume c) Organization Health
a) Number of trained 11. Budgeting and Accounting : RoI, Return per
employees employee
mance
Pay Rate Differential between
high performers and others
High Performers Turnover
Rate
% Across the 9box grid
 HR STRATEGY MAP
Improve profit
margin
cial
Maximize return Minimize cost
ective spent
per employee

Opportunity to Efficient
Build technical explore
mer employee
competencies advanced levels
ective utilization

Knowledge Streamline cal

Technical Internal
ss management trainings promotions
handling
ective process

gic Talent
ctive development Customer Performance
oriented service Increase
culture
productivity
 PERFORMANCE MANAGEMENT FRAMEWORK
Objective Performance Weightage Maximize Minimize Enabling factor
measure Human HR cost
capital
Improve profit Net income per 15% Aligns employee goals to business
margin employee objectives

Maximize return ROI per employee 10% Provide clarity on individual goals
per employee develop individual action plan

Minimize cost per HR cost per employee 10% Promote effective utilization throug
employee engaging employees

Build technical Level of competencies 15% Impart training based on training n

competencies accomplished
Opportunity to Number of internal 10% Promote internally based on emplo
explore advanced promotions potential
levels
Efficient employee Adherence to 10% Establish systematic processes to
utilization schedule handle customer queries
Objective Performance measure Weightage Maximize Minimize Enabling factors
Human capital HR cost

Knowledge Number of best practices 5% Builds tacit knowledge within the functi
management shared with peers and collaboration among peers

Technical trainings Number of trainings 7% Improves quality of recommended cust

completed solutions

nternal promotions Number of promotions 5% Promotes internal culture along with ta

development

Streamline call Adherence to call duration 8% Benchmark call duration based on sev
handling process issues

Talent development Talent gaps identified 5% Helps in developing action plan to bridg
identified gaps through buying or buildi
talent

Customer oriented Customer satisfaction 8% Maintains excellence in customer servi

service score

Performance culture Number of top performers 5% Develop competitive spirit through

incentivizing or rewarding high perform

ncrease productivity Number of customers 7% Provide timely feedback on employee

satisfied with employee performance
recommendations
Identifying areas in the HR life cycle where
advanced analytical techniques can be
leveraged.
Learning the application of various analytical
techniques across Talent Acquisition.
Recruitment Analytics: Case study at Monoce
 hich Frameworks are a must have?
Cost-Benefit
Benefit Analysis

• Capital Budgeting • Guestimates • Break-even

• NPV/IRR Analysis • Scalability • Sustainabili

milar templates on: Lucidchart.com/Journey-maps

ey Performance Indicators Data figures

Acquisition Rates Quotes and Dollars

NPS
Upsell/Cross-sell rates
Lead counts
ecruitment Analytics Journey at Monocept
e Recruitment Initiative How we did it

Recruitment Process Outsourcing vs Internal Cost-benefit analysis was used to arrive at the
conclusion that we should build a inhouse Recruitm
instead of outsourcing

Feedback Approach based on high rejection ratio Recruitment data and yield was captured and was u
to come up with attending every interview and start
capturing those questions
Recruitment Drive vs Weekday interviews hypothesis It was found that more candidates are available dur
testing drives than weekday interviews.
Turbohire a new platform was built keeping in mind Ability track the time, yield and timeline
Monocept analytics interests
Behavioral insights and test Text Analytics using IBM Watson
Online coding test vs Inperson test, Scope vs Identified Method1-Method2 Test Positive score.
Scalability
Breakthrough/Bottleneck analysis Flow increased 3 times
Prescriptive Analytics of Recruitment Identifying the Parameters which can reduce the
resumes processed
Learning the application of various analytical
techniques across Talent Management.
Dimensionality Reduction
Feature Selection: Extracting a subset of features that have
maximal impact
Feature Extraction/Reduction: Using a linear combination of
features so as to explain a majority of the variance
Principal Component Analysis (PCA)
https://www.youtube.com/watch?v=HMOI_lkzW08
Clustering & Segmentation
a distance measure
a clustering algorithm
the distance between two clusters
mine the number of clusters
te the analysis
m is to divide the population into groups
s similarity / dissimilarity?
.NET
o you define distance between two clusters?
Technology

Finance
High
JAVA
Performance
Sales
10000 Medium
Employees Performance
Marketing
Low
Performance
 Time series Analysis

What do you make sense of the followin

Source: analyticsvidhya.com
Model Evaluation - Truth Table
ACTUAL • SENSITIVITY
+ - • TP/(TP+FN)
• SPECIFICITY
• TN/(FP+TN)
+ TP FP • POSITIVE PREDICTIVE VALUE
• TP/(TP+FP)
• NEGATIVE PREDICTIVE VALUE
• TN/(TN+FN)

- FN TN
• ACCURACY
• (TP+TN)/(TP+FP+TN+FN)
ment Analysis
Text Analysis based on questionnaire: IBM Watson
• https://personality-insights-
demo.ng.bluemix.net/
Organization Network Analysis
y organization, people build informal “go-to”
to” teams. They rely on that one person who always knows “how we do things h
nd someone in finance who can answer any budget question. These spontaneous, critically important connections are th
d of organizations worldwide. Organizational Network Analysis (ONA) is a structured way to visualize how communicatio
tion, and decisions flow through an organization. Organizational networks consist of nodes and ties, the foundation for
fo
tanding how information in your organization is flowing, can flow, and should flow.
ONA - Methodology
The questionnaire was not anonymous. This was compensated by making
it more informal, circulating it through whatsapp and using informal text
like smileys, three dots “…”, etc
The questionnaire addressed two aspects – “to whom” the student asked
the questions listed, “from whom” such questions were asked to the
student. This was to identify cliques in the network.
The questions were categorized into formal (career-related,
(career academics)
and informal (daily
daily gossips and chats) questions and were jumbled to
maintain the informal style of the questionnaire.
Learnings : We tried to analyze how different cliques and network patterns
affected organization performance.
ONA: Glossary
Node: Sometimes called a “vector” because it marks the • Ego network focused around one central node. A
intersection of lines, and sometimes called an actor, example is a correspondence network derived from
nodes are the elements of a network that are being collected letters of a single individual.
connected.
Edge: Connection, link, or tie between nodes. •Conductance – Measure of how well-knit the grap

Basic Network Measures •Transitivity – Extent to which relation between two

connected by an edge is transitive
degree of a node: The number of edges connected to
this node. Variants include in-degree/out-degree, which
•Mean Distance – Mean of distance between all pa
counts the number of ingoing and outgoing edges in a
directed network. Sometimes indicated by the size of the
sphere representing the node. Also called degree •Edge Density – Number of edges / maximum poss
centrality. edges

Clustering Coefficient: •Network Diameter – Distance between most dista

Eigenvector centrality: Eigenvector centrality
•Structural Holes
measures the “influence of a particular node . . . by the
connectedness of its closest neighbors. This can be
thought of as the who you know type of centrality, •Betweeness Centrality
wherein an individual node might not be thought of as
important on its own, but its relationship to other highly •A clique in the social sciences, is a group of individ
connected nodes indicates a high level of influence” interact with one another and share similar interests
ONA: Glossary
ode: These are the people who seem to know everyone.
des share lots of information and influence groups quickly.
des can be anywhere in the hierarchy of an organization,
well liked, and are highly engaged in company news and
ents.
e broker: These people create bridges between groups.
owledge brokers, information and idea sharing grinds to a

l: Easily overlooked and unconnected to the rest of the

high-potential
potential peripherals can be a risk to organizations.
al Java coders who don’t teach others best practices not
ate product development, they are also easily convinced to
alents elsewhere.
are the formal and informal relationships between nodes.
ng optimal relational ties between central nodes and
brokers helps ensure useful information moves easily
nd within groups.
Survival Analysis
mate time-to-event
event for a group of individuals, such as time until attrition
ompare time-to-event
event between two or more groups, high vs medium vs low salaries
ssess the relationship of co-variables to time-to-event,
event, such as: does salary, performance, promotions
ence time to attrition of employees?
: expected time-to-event = 1/incidence rate
metric and Non-Parametric
Parametric: Exponential
Non-Parametric: Kaplan Meier
Introduction to Kaplan-Meier
Meier
Non-parametric
parametric estimate of the survival function.
Commonly used to describe survivorship of study population/s.
Commonly used to compare two study populations.
Intuitive graphical presentation.
Survival Data (right-censored)
Employee A
Employee B
Employee C

Employee D
X 1. Employee E
Employee E leaves at 4 months

Beginning of study  Time in months  End of study

Corresponding Kaplan--Meier Curve
%

Probability of surviving
to 4 months is 100% =
5/5

Employee E leaves at 4 Fraction surviving

months this death = 4/5

 Time in months 
 Survival Data
Employee A

Employee B 2. Employee A status

unknown after 6 months

Employee C
4. Employees B and
D survive for the
Employee D 3. Employee C leaves at whole year-long
X 7 months
study period

Employee E

X 1. Employee E leaves at 4
months

Beginning of study  Time in months  End of study

All this is fine but how do we learn People
Analytics in College, especially studying from
home?
1. Use analytics as a part of every project you prepare
2. Consider college is an social organization of about 300-500
300 members
3. Use sample data sets from Kaggle and other sites and start using these sets for
doing your analysis
4. Frame your hypothesis, correlations and keep testing them.
5. Use prepared models and test the models through statistics to check their
elevance.
essment Centre- Visualizations
Informal Network Analysis

ossip Graph of PGDHRM 2017-19 batch based on the responses Figure 2. Gossip Cliques of PGDHRM 2017-19 batch based on the resp
Knowledge Network Analysis – Cont.

cept explainers of PGDHRM 2017-19 batch based on the responses Figure 4. Maximum Informal Connections Network of of PGDHRM 2017-1
on the responses
Informal Network Analysis – Cont.

Figure 5. Maximum Formal Connections Network of of PGDHRM 2017-19

2017 batch based on the responses
hank you for listening through!