How to do grouping of analytics data?

How do I get started in HR predictive analytics?

• Any practical guide that you can share? Or step by step? Hi, I am practically still trying to get HR domain experience, but still far from it. And I want to get past the usual Descriptive & Exploratory Analysis that we already have and just start doing Predictive Analytics. There's a lot of info out there on the benefits, solutions, broad steps, questions/answers,etc but no actual basic steps to do it. I understand it doesn't fit all organizations and data, or it depends on what questions you want answered, so here's the sample data that we can use.... NAME BIRTH_DATE GENDER CIVIL_STATUS POSTITION RANK RANK_TYPE (officer,staff,etc) GROUP_NAME (organisational grouping) SUBGROUP_NAME (organisational grouping) UNIT_NAME (organisational grouping) SUPERVISOR_NAME SPEC_AREA (operation,support,sales,etc) SCHOOL DEGREE HIGHEST_EDUC EMP_STATUS (probationary,regular,etc) DATE_HIRED RESIGNATION_DATE REGULARIZATION_DATE REGULARIZATION_RATING So what basic Predictive Analytics can you derive from these variables? Edit: As per the recent posters, I did mention that of course it will depend the questions you want answered. (sorry if I can't convey that enough) Let's just say that HR can't explain appropriately to IT want they really want like (I want a what-if scenario if I increase Benefit A can you predict attrition decrease by 6 months on these classifiers? something like that.). Let's start from the very basic please. Another way of looking at it is for example you already gave them a system for descriptive & explanatory analysis. Now you want to impress them further by starting on predictive analytics on your own by analysing the data above. (it may not solve a business problem that they would appreciate but at least its a start or make them realize that they could start doing that already) Or to put it in another way. what's the low hanging fruit here? tia

• Answer:

  Predictive analytics means you use past data to make decision that impacts the future outcome. So what is the decision you are trying to make and what future outcome do you want?  Is this for recruiting? Employee evaluation? Employee turnover? You need to define the problem in an appropriate way, then find appropriate data to model the problem. Not sure which of the variables in your sample data will be useful without knowing the problem to solve.