Utilization of Exploratory in HR Business

2019/07/29(金)に開催したExploratory データサイエンス勉強会#10の株式会社JTBコミュニケーションデザイン様のご登壇資料です。

Transcript

1. Utilization of Exploratory
   in HR Business
   OHIRA Yusuke
   

   View Slide

2. Who?
   • Name : OHIRA Yusuke
   • Age : 35 years old
   • Employed : JTB Communication Design Inc.
   • Job : Citizen Data Scientist
   • Domain : Human Resource Business
   • Used Tool : Exploratory, Microsoft Power BI
   • Community : Data Science Study Group,
   Tokyo.R Study Group, Power BI Study Group
   

   View Slide

3. Joined the Data Science Boot
   Camp about a Year Ago.
   

   View Slide

4. Today’s Target
   Those who want to know how to use
   Exploratory in various domains.
   Those who want to start HR Analytics
   as a company or an individual.
   

   View Slide

5. Analytics in the HR Domain
   People Analytics
   

   View Slide

6. What is
   People
   Analytics?
   Analyze the Data of
   Employee Attitudes, Behavior
   and Outcomes.
   Executives and Management
   make Decisions based on
   Objective Indicators.
   

   View Slide

7. Drive
   Traditional
   HR
   •Intuition
   •Judgement Calls
   •Instincts
   •Experience
   

   View Slide

8. People Analytics is Important
   86%
   

   View Slide

9. Capability of
   People Analytics
   • Workforce Planning
   • Sourcing
   • Onboarding
   • Engagement
   • Human Development
   • Internal Mobility
   • Retention
   • Wellness, Health, Safety
   

   View Slide

10. Start with…
    Data? Business Question?
    

    View Slide

11. IMPACT
    Cycle for
    People
    Analytics
    Identify the
    Question
    Master the
    Data
    Provide the
    Meaning
    Act on the
    Findings
    Communicate
    Insight
    Track the
    Outcome
    

    View Slide

12. PEOPLE ANALYTICS IS
    JOURNEY WITHOUT TERMINAL
    

    View Slide

13. Identify the Question
    HR Activity
    & Process
    HR Outcome Organizational
    Objective
    

    View Slide

14. Four V
    Required For
    “Big Data”
    • Velocity
    • Variety
    • Volume
    • Value
    

    View Slide

15. Does the Data have Value?
    Assumption
    Hypothesis
    Theory
    

    View Slide

16. Guess Components
    Outcome
    Component
    A
    Component
    A-1
    Component
    A-2
    Component
    B
    Component
    B-1
    

    View Slide

17. Issue of Implement
    People Analytics in Japan
    No
    Database
    Data Siro,
    Data
    Governance
    Dirty Data
    Decision
    Making
    Culture
    

    View Slide

18. Data Preparation and
    Data wrangling is
    90%
    

    View Slide

19. Employee
    Survey
    As a Initial Process in
    People Analytics
    • Get the data that can be
    analyzed for now.
    • Increase the certainty of the
    Assumption.
    • Approaching a Causal
    Relationship.
    

    View Slide

20. Analytics by Exploratory
    Design Questions
    based on Assumption.
    Regression Analysis
    verifies relationship
    between outcome and
    other components.
    Explore relationships
    between various
    components by PCA.
    Observe trends in
    Free-text by NLP
    

    View Slide

21. Take Action!
    • Fear of “And then?”
    • Start with a frontline worker.
    • Actions are part of the analysis.
    

    View Slide

22. Causal Inference Based on
    Static Dataset
    Static
    Dataset
    Analytics
    Causal
    Inference
    

    View Slide

23. y = ax + b
    Outflow Inflow
    

    View Slide

24. Causality Problems
    • Omitted Variable Bias
    • Reverse Causality
    

    View Slide

25. Approaching a Causal
    Relationship
    • A Business dose not have just
    one issue.
    • The cause of change after
    intervention is not always that
    intervention.
    • Compare the intervention
    target and non-target.
    

    View Slide

26. Familiar is Paramount
    

    View Slide

27. Data
    Scientist’s
    work is quite
    divers
    Identify the
    Question
    Master the
    Data
    Provide the
    Meaning
    Act on the
    Findings
    Communicate
    Insight
    Track the
    Outcome
    

    View Slide