Driving the Future of HR with Analytics and Bots
- 1. 1 Data Analytics – HR 2020
- 2. 2
- 3. Introduction Breaking AI into HR Context Overview of Data Analysis Use case of AI – ML, NLP, Image Processing and Sentiment Analysis Flow of data from unstructured sources How doe we convert Unsupervised data to Supervised Information Case study from Real time implementation of Analytics How BOTS and Analytics is the new norm for 2020 + How do we use AI - Libraries in HR context Questions and Answers What will we cover 3
- 4. 4 Let me Start with a Short Story …..
- 5. Classification Problems • “Will my candidate join - The candidate was excited” • “I thought the JD was correct but there seem to be a problem with the real work against the JD “ • “Do I have a Heat Check method to gauze the joining probability” • “Are my employees Engaged?” • “How can I increase my Training Participation” • “Can I know my capacity numbers before hand” • “Are my assumptions and Hypothesis correct and mapped with exact results” • “Is My succession Planning method adding to the business cause of Cost and Returns “ • “Can I wear the HAT of CXO” Do I have the right model • Effective Model to Describe Data flow • Effective Model to Predict Performance. • Effective Model to Predict Trends Types of Problems we face in our daily life….. 5
- 6. HR is Beyond Statistics …………… 6 Should I have Mastered Statistics rather than being in HR? The Subject of HR is Very Different in different industry Context Why can’t management distinguish Administration and the subject of HR
- 7. Let’s See how can we break Analytics into Simple Steps 7
- 8. Stages of Analytics 8 Graphs , Visualizations helps us understand the various trends in the HR. Performance Trend , Leaves Pattern , Attrition Trends Important as it gives a deep dive and provide us with insights or reasons for various trends. If the present conditions carries on then what might happen or what it could lead to if the current process follows on. Recommends possible course of actions while guiding to a solution.
- 9. Let’s Begin with the concept of HR Analytics.. 9
- 10. Some technicalities of Analytics ….. 10
- 11. Assumptions : 1. Linear Relationship 2. Multivariate Normality 3. No Multi- Collinearity 4. Homoscedasticity • Y the dependent variable. • X1 , X2 , X3 are the various independent variables on which the value of y depends. • M1, M2, M3 the coefficients of dependent variables. • b the y intercept or the error term. • We use a sample data what we called the training data set. Try to predict or find the values of the above formula. • Accuracy is checked by the “r squared” value. If r2>0.75 we can say it model is good. The higher the r square the more accurate is my model for predicting Y. • Significance of X variables can be checked by the p value . If the p value <= 0.05 the variables are considered very significant for the model. • Now, once we get the values on the above formula and the accuracy is to our liking we can use the model (above formula ) to predict Y for the population. Simplifying HR through Statistics 11
- 12. Usage of algorithm like : • Logistics Regression , Naïve Bayes , Decision tree , Random Forest which ever model shows a good accuracy can be used to predict classification. Could be used to predict attrition etc. • The accuracy of the model for classification models is checked by the Confusion Matrix. Simplifying HR through Statistics 12 Accuracy : TP+TN / (TP+FP+FN+TN)
- 13. 13 Use case context of data flow by understanding AI from an HR Angle
- 14. Integrated View - HRMS, Micro, BOTs & Analytics 14 » Reduce Recruitment TAT through ML and NLP capability in JD creation, Sourcing, Assessment and Engagement » Create digital directories of candidates and employees » Implement automated ‘Web Check-in’ process for Onboarding » Link employee to a ‘Drag- n-Drop’ Org chart » Track assets assigned to employee » Define and initiate Early 15-30-45-60-90 days feedback » Digitally create Employee Road Map - 365 days and beyond » Create Confirmation Assessment Profile » Employees’ View of documents, letter, policies and forms » Managers’ view of employee records » Geo Tagging and Fencing for Attendance Management » Capability to launch Interactive bots » Enable employee to access from both web and mobile » Conduct Employee Stay Interview Process » Conduct 360 and Skip to understand supervisor effectiveness » Conduct Behavior Analysis through digital engagement (incl. BOTs) » Identify Training Needs and create Career Progression road map » Conduct Assessment Center LMS » Assign learning Modules » Publish training calendar & Assessment centers » Integrate Digital Library with open content through API » Design Instructor Led Workshops » Launch Survey, Quizzes and Feedback » Dynamic reporting » Conduct Webcast and Webinars Exit Management » E-Exit BOT to provide discussion points to HR » Exit Bot for Paperless Employee Exit » Online Departmental Clearance Integrated HRMS Platform for end-to-end employee Life Cycle Management Employee Directory (ESS) Recruiting & Digital Onboarding Attendance & Leave Mgmt. Employee Mgmt. Feedback & Retention Performance Evaluation Learning Management System Employee Exit
- 15. How AI and ML will be the part of the Entire Life Cycle of Employee to Simplify 15 Pre Joining Data set 1 AI based Sourcing ML based CV Analysis Competency Based Profiling Gap Mitigation against JD Profiling Profiling Intimation to all Department Joined Data Set 2 Transition – Pre Joning Induction 30 Days Data Set 3 Early Stage Stay Interview LMS Governance Knowledge Transfer 60 Days Data Set 4 Early Stage Stay Interview LMS SKIP Knowledge Transfer Performance 90 Days Data Set 5 Early Stage Stay Interview LMS SKIP Knowledge Transfer Performance Profiling Gap Mitigation 180 Days Data Set 6 Early Stage Stay Interview LMS SKIP Knowledge Transfer Performance Profiling Gap Mitigation Building Road Map Conversion of Unsupervised data to Supervised Data in ML FOR Predictive profile PREDICTABILITYPROFILE
- 16. 16 Lets look at the real context Step 1 Step 2
- 17. Step 1 for Analytics - How AI and ML will work? 17 TA posts a position on Career/Opening Portal TA Receives Job Application/ Resume TA Analyze the CV and compare with the required JD Documents TA finds the suitability Thank you for your interest No Candidate receives an email for first round of Interview/ screening Yes Interview gets Scheduled An info email goes to Interview panelist Interview took place and panelist published the scores/feedback Selection Status Negative/Disqualified BOT Process Positive / Qualified QR Code gets generated and information goes to SPRINGBOARDManual + Web Based input TA Team · BGVC Check · Profile Check · JD ANalysis Supervised Data Set Machine Learning Game Changer – NLP and ML and Indexing
- 18. Step 2 - Flow of Employee data -1 – 180 18
- 19. 19 Use case– Descriptive and Predictive HR Data and the Game Changers https://www.stratemis.com/analytics-as-a-business
- 20. Recruitment Applicant Tracking – Descriptive Dashboard 20 Lets Build the JD through ML and NLP How do we use NLP Libraries Support Prescriptive Analytics https://hris-demo.stratemis.com/ess
- 21. Probability of his joining the company / organization Bases on the positive polarity scores Recruitment Applicant Tracking– Predictive Dashboard
- 22. • 5 Engagement Positive Polarity Scores of the candidates • Positive Polarity would give us a more promising figure to know “How Positive he is in joining.” • Polarity out of 20 conversational points . • We have used Sentiment Analysis on his responses to calculate the positive polarity. Recruitment Applicant Tracking– Mechanics
- 23. 30 % weightage from Engagement 1 30 % weightage from Engagement 2 20 % weightage from Engagement 3 10 % weightage from Engagement 4 10 % weightage from Engagement 5
- 24. Probability of his joining the company / organization Bases on the positive polarity scores
- 25. Head Count and Attrition Report – Descriptive 25
- 26. 26 Head Count and Attrition Report – Predictive https://hris-demo.stratemis.com/ess
- 27. 27 Use case HR – How do we reach to a Hypothesis To Predict Attrition https://www.stratemis.com/analytics-as-a-business
- 28. Attributes to Attrition or Performance - Influencers 28 Attributes 1 3 5 One on One assessment - Concerns-Redeployment/ Night shift/ Family problem/ Outstn FTE/ Training Issues / Marriage / Salary/ Higher education/ Others (If others, substantiate the reason on comments section) Any of the mentioned concerns raised repeatedly/ still to be addressed Raises concern but concern addressed No concerns raised/ no sign Performance Visible dip in performance along with concerns on other stated attributes. Overall consistent performance, however with a few sporadic instances of low quality. Highly motivated and good performer. Leave Long leave/ Dubious leave/ Unplanned / Unscheduled leave Leave extended (Short or long), not planned in advance (appears genuine) No Unsched Leave taken External Interviews Heard from the associate himself/ herself Have a haunch (word on the floor)/Confirmed from other sources No sign Behaviour/ Motivation Energy drop/spreads negative attitude Indifference towards process. Close to associates with negative energy. Not open to changes Low energy/ tends towards negative attitude but can be addressed through pep talk No participation in process related extra activities. Does not have a negative approach to work but needs to b Extremely high energy, positive attitude Willingness to take on higher responsibility. Takes initiatives and shows the crave for learning more. Has the right appraoch Personal Effectiveness Has not taken any initiatives with regards to process improvement. Does not believe in team work. Has major issues with reporting manager. Has potential for additional responsibilities , needs to be pushed & delegated. Only Meets the performance criterion. Extensively involved in projects , additional responsibilities & keen to take On assignments. Got promoted within last year. Helps the manager Career Growth Has been in the same role for more than 2 years and strongly feels that Career growth is over-due Has been in the same role for more than 18 months and has indicated that Career growth is due No Issues Skill set Skill set Mismatch New to the industry and mentioned the difficulty in understanding the process, once the employee is on the floor post successful completion of OJT. Has experince in the industry but needs coaching and supervision. Is taking time to come up the learning curve No Skill Issues SIPOC Question before Modelling 1. Who is providing the Data ? 2. Who is processing the data ? 3. Who is benefitting from the data ?
- 29. Attributes to Attrition or Performance - Contributors 29 Reference Guide for Attribute no. 1 (as stated above) One on One assessment - Concerns- Redeployment/ Night shift/ Family problem/ Outstn FTE/ Training Issues / Marriage / Salary/ Higher education/ Others (If others, substantiate the reason on comments section) 1 3 5 Redeployment ( to other process / business) or Relocation ( to other location) Actively seeking redeployment / difficulty in redeploying Has talked about redeployment though not pursuing actively No issues Marriage in offering Would definitely leave after marrriage Might leave… No issues Night shift Has health problems due to night shifts…asked persistently for a day shift Has mentioned about moving to day shift…not persistent No issues Family problems We know these are causing problems at job…performance, late to work, etc Seems to be able to cope No issues Salary Extremely dissatisfied has been talking about it on a regular basis. Was dissatisfied, after discussion seemed satisfied and has not raised it again No issues Higher education Studying for entrance exam, might leave once he/she clears the same Mentioned interest in higher education…might be applying for the same No issues Medical problems Severe problems interfering with work Problems…able to cope with difficulty No issues Warnings - diciplinary, behavioural Given more than one warning on discipline /behaviours to be followed Given warning on discipline /behaviours to be followed and shows good improvement No issues Maternity Asking for frequent leaves or too many last minute leaves in the initial period of maternity. Staying alone with husband & worried about lack of support at home in terms of Child care. No issues
- 30. Legends to Determine the Output 30 Legend 1 High Risk Red - Attrition anticipated in the next 1 month, immediate attrition (we know that the employee has gone for an interview, told someone that he/she is going to leave, patterns in absenteeism that reflect poor engagement, or any other reason etc) 3 Medium Risk Yellow - Attrition anticipated in the next 30-90 days. Factors that can lead to the employee leaving the job in near future but don’t expect it to be immediate. (Slow but steady signs of poor engagement, lacking attention to details, lack of ownership, increased error rate etc., contrary to the individual's usual self) 5 Low Risk Green - Attrition not anticipated, required levels of engagement seen.
- 31. Supporting Enablers for HR to derive to a Attrition Analysis – Prediction 31 BOTS Engagement Data Demographic Data Supervisor Data Individual Engagement Data 360 Feedback Stay Interview Leaves (Absenteeism) Performance Motivation measure check Learning Agility • Compensation – Competition Compa ratio and Performance Tertile • Supervisor and Mentor
- 35. 35 Attributes to Attrition or Performance and Behavior Influencers BOX Model
- 36. To drive Analytics, HR need not be Statisticians….. 36 • Who is providing the Data ? • Who is processing the data ? • Who is benefitting from the data ?
- 37. Our Digital Solutions & Enablers 37
- 38. About Stratemis 38 The only SaaS based HRMS technology provider with Stratemis is an agile and intelligent HRMS platform, focused at making HR processes intuitive for operating managers and employees 1. 12 Core modules and 2. 12 Micro-service modules Enriched by: A. ML based BOTs and B. Intuitive and Predictive Analytics Our platforms bring together all constituents – Employees, HR Managers, Operations Managers – on to a single, easy to use and intuitive operating environment The only SaaS based HRMS platform providing complete Employee Life Cycle Management
- 39. Stratemis’ Promise and Commitment 39 Stratemis’ Beliefs Pace Passion Transformation Stratemis’ Purpose Help customers achieve HR digitalization with marginal cost & minimal disruption Stratemis’ Promise To Provide best in class AI driven modular HRMS framework Stratemis’ Commitments Meet customers’ HRMS requirements through a single set of platforms. Achieve customers’ HR Digitalization objectives through Configurable and Modular platforms Help implement highly configurable and agile HR processes across all industries Minimize Customers’ CapEx Ensure smooth implementation through our published API
- 40. 40 Thank You!
- 41. 41
- 42. 42