The End of H.R. Theory

Back in 2008, Chris Anderson, the Managing Editor of the technology-focused magazine Wired, wrote an article titled The End of Theory: The Data Deluge Makes the Scientific Method Obsolete. In it, he argued that we need to relook at the way science is done, because the ‘old’ way of doing science, in essence, a system where we built models and tested them to find out ‘why’ things happen in a certain manner, is irrelevant in the contemporary context where data is abundant. Using the power of Big Data and smart analysis, he argued, we no longer need to hold fast to the adage that ‘Correlation is not causation’ – correlation, he argues, tells us when certain things will happen, and that’s all we really need.

Today, I’m calling a similar end – the end of H.R. theory. Gone are the days and ways of old, when we would all sit around speculating why certain people leave organisations, and what to do about it. With the ever-increasing availability of data, all we need are smart tools that can help us understand exactly when a person is likely to leave an organisation, and what can be done to prevent this from happening.

But there’s a catch in there – as there always is! Data exists – but it needs to be captured. H.R. today needs a culture of counting everything – and the only way to do this is by using a powerful set of technological tools, such as those that come together to form IDfy’s incredibly intelligent H.R. services platform. Pen-and-paper, simply put, just isn’t going to cut it in today’s world.
The second step one needs to take is building the ability to understand and make sense of all the data we’re collecting. This requires an aggregation of abilities across disciplines – data science, technology, H.R. and managerial expertise, and statistics. Unless you have these, all that your Big Data does is take up storage space, wearing a forlorn expression, hoping that someone will actually get around to using it soon. I have the incredible luck today of working with just such a team at IDfy today.
Finally, you need to draw insights from all that analysis – what are the parameters we should be testing? On how many counts should we do correlation and regression analysis? What combination of questions yields the answers that will actually help our work? This is where Deep Learning concepts such as Supervised Learning, Unsupervised Learning, and Reinforcement Learning need to be applied in tandem, so that we obtain benefits from both, the questions we ask (Supervised Learning) and the questions we don’t even know we need to ask (Unsupervised Learning, similar to ‘Data Mining’), so we have a system that gives us the answers we need (Reinforcement Learning). All of which, of course, constitute Deep Learning.
Here’s a tiny example: very often, recruiters are concerned that the candidates they’re considering for a particular position may not have the qualifications they say they do (by the way, this happens more often than you would think!) Let’s say that there are nine parameters that people could lie about in this regard:
  • Course completion date
  • The name of their degree
  • Their roll / registration / seat number
  • Their specialisation or major
  • Their grade / C.G.P.A. / percentage / final score
  • The type of course
  • The institute they studied at
  • The certifying authority
  • The type of degree

If you’re suspicious about a candidate, you could, of course, examine each of these aspects in minute detail – but wouldn’t it be more worthwhile to focus on the aspect that you are most concerned about – or even better, the aspect that a candidate is most likely to lie about? Here’s a picture that shows us exactly what these areas are:

EdQualFalsReps

 

Now, imagine what you could do by examining correlations against some more parameters – what degrees or qualifications attract the most number of false representations? From which university are there most likely to be the most number of false representations? From which city? Which age group – freshers, or laterals?
Insights like these help you understand where to look, and how to improve the accuracy of your checks – and if you’re partnering with an organisation like IDfy, which can bring you the combined expertise in technology and data science that is so vital to such an exercise, you’re all set to call an end to theory – and a new beginning of certainty!

 

Leave a Reply

Your email address will not be published. Required fields are marked *

Show Buttons
Share On Facebook
Share On Twitter
Hide Buttons