I have been blogging on and off on the uses for big data, and most of my examples come from the manufacturing / asset management world. For example: predicting equipment performance, product quality. Just mining the data that chemical companies have been collecting forever. These to me are obvious uses for analysis and use of big data.
I was reading this blog on Workforce talking about using data and predictive analysis to identify a candidates potential to perform in the role being hired. This really intrigued me as an idea, after all most companies have some policy where there is a probation period, so if the candidate does not work out they can be let go or returned to their previous job with no penalty. But this still consumes a lot of time and over the period wastes a lot of resources (e.g. on boarding, training, etc.). So if we could more reliably predict the likelihood of a successful candidate all would benefit.
One key line in the blog "the size of the data set is critical, quality of data is important as well". This seems obvious, but are organizations collecting data in such a way that it can be used to predict the success of a potential employee. And if so what type of data are they collecting;
Employment History; I am sure that my current company does not have my complete history in a usable form. In fact I would be surprised if they even had my original application and resume (after all it has been almost 20 years since I joined SAP). So they would have (could have) a history of the work I have done while at SAP, but would this be enough detail to generate a prediction of success? Do they have this information for all their employees who are successful in similar positions, do they have information on employees who were not successful in those types of positions?
Skill sets: It is fairly easy to list out your skill sets, but how do you rank your ability (high, medium, low?). In addition keeping the skill sets up to date is always a problem. Internal training can generate automatic updates to a skills data base, but what about training, activities that the candidate acquires outside of work? How can you deal with partial skill sets? Can master in one skill predict your success in mastering another?
Role / Job Definition: Is the job defined in enough detail that we could actually predict success? A lot depends on the job description. I know how easy it is to create a job description that only one candidate could fill all the requirements. So, in my opinion there would have to be some standardization on how to build out the job requirements.
What is a successful candidate? A stupid question, maybe. But is success defined for example as a) being able to do the job satisfactorily, b) exceeding expectations, c) staying in the job a set length of time, (1 year , 2 years, etc.), d) being a candidate for promotion? Would you accept a candidate who was predicted being able to do the job extremely well, but would leave after 1 year? Or would you accept a candidate who performs satisfactorily and will work for you until they retire.
As I said above the concept of using predictive analytic for talent acquisition intrigues me, but I think that a lot of work needs to be done around the topic before it becomes common place. Nice idea though.
