The word data science did not mean too much if you go back a decade; at least not to people who had nothing to do with the academic pursuit of the discipline. The world has truly changed over the last decade and the commercial environment is increasingly data centric. We have been reading a lot of arguments about the development of AI has suddenly disrupted the job markets and changed the industrial sectors. Well, has the world really changed so fast?

If you dig in, you will find that AI as a technology was researched and developed much earlier and machine learning was also developed as the core of AI technology. Why then has it seems to have erupted all of a sudden? Funnily enough, before machine learning could change anything it has been changed by the world – because it, the world, changes fast.

Need data? You have it now

Machine Learning feeds on data. More the amount of correctly put data, higher the possibilities of getting desired results. You do not already know how the method works; it is basically all about training a computer to learn by itself – to recognize patterns without human intervention. This sounds way simpler than it is in reality. Once you have decided what you want to achieve with the process, which is not a walk in the park either keeping the stakes in mind, you need to set the algorithms. The machine now follows these algorithms to recognize the difference between two pieces of unstructured data. The hardest part starts right there. The algorithm needs to be tested against prototypical data mined especially for that purpose.

Now, the main reason why machine learning technology did not take the stage earlier is that there was not enough data available to test the machine learning algorithms. Now, we all know, if there is something that is available in abundance, it is data. Companies have access to unbridled data – most of it being unstructured, textual and very random – which they can use to get into the head of the customer. And, luckily, the computational capabilities have been developed to a certain level that can handle these tasks.

What, then, seems to be the problem?

If you are willing to find a problem, one will definitely show itself. There are two problems regarding the matter we are engaging ourselves in:

• The lack of skilled professionals
• The fluidity of the market

The first one seems to be very simple. Machine learning is tough stuff to learn; let there be no doubts about that. In the Indian context there is a lively attempt by universities and private institutes alike to fight the skill shortage. You can easily find a machine learning course in Gurgaon or Bangalore or Hyderabad in many cities in fact.

The second problem however, is more complex. The machine learning methods are modeled with the help of data. This data takes a lot of time to be prepared and the deployment costs money – a fair amount of it. Once in a while these algorithms need to be re-evaluated and re-modeled; basically because of the changing nature of the world. Some say that adapting DevOps workflow might help the situation, but the field is young and you never know where it goes. One thing is for sure that there are a lot of vacancies. Get yourself ready with a machine learning course.