Saying NO to Machine Learning !!!

Now this should catchy and make you wait for a while. Machine learning or ML, which has been the most happening buzzword is looked upon as next revolution post Internet in area of computer technologies by many.Like as Internet was considered and has been disruptive to many legacy procedures and convectional thought process, Machine learning and Artificial Intelligence (AI) are expected to re-write the same story.

Modern markets and corporate world are driven by prime focus of improving the bottom lines. In nutshell they breathe profitability. It has been learnt that companies and organization which did not adapted the change in technology eventually ended up fighting for their existence in fiercely competitive market, no matter how much impact they had on industry and society . Hence the technical torch bearers of all organizations are under constant dilemma that whether adopting or investing in to a new technology would actually create any value to shareholders/promoters of the organizations.

The write up here is maiden attempt for responding towards contrasting subject under lens. It will tend to unearth those areas which are generally taken as accepted while riding the wave of buzzing technologies around. In absence of proper introspection, the results do not fall in line having adverse effects on prime elements i.e social, human and business. The following are the aspects that needs to be considered before we turn on the green light to start throttling on ML&AI.

Unsure on Problem Statement: When is comes to ML, problem statement is the Numero Uno aspect to watch out for. Irrespective of what resources are at the disposal, the problem statement should be absolutely clear. This would not only assist to assess the idea, that what needs to be done to achieve the goal that is set, but would also lead to digging out multiple way outs to solve the problem, of which ML can be one of the method. The other aspect i.e unclear problem statement is exactly analogous to situation of design imperfection in a monument.Not having absolute clarity on problem statement in business sector can lead to substantial capital loss.

Not having Required Data : It is now a well know fact that we train machines in the same way as we humans do i.e by Experiences. Going Technically, experiences form the training data set for any model and henceforth the predictive analysis. With efficiency in computing and storage technologies, organizations are now able to collect data at scale. For organizations which do not emphasis on data collection, there is no space available in this area.Hence it is obvious that there is an immediate need to setup facilities and develop ecosystem to collect data.

Undisciplined Data Collection Process: Considering for an organization if above mentioned requirements are met, the process for stepping up the stone for Machine Learning is not ready yet. For process/organization, where data collection is not aligned with mission of the organization will ultimately lead the organization to loss making quadrant. Hence having structured data has equivalent mandate of collecting it. Misalignment with the mentioned has all the might to bring the famous saying in data science come true i.e "Garbage IN , Garbage OUT"

Not Prepared to Accept Failure of ML Model: One of the famous proverbs "Learning is never ending process", holds true for machine learning models as well. It implies that machine learning is an iterative process wherein the output of test data set is fed back to experiences to make the model more precise. Having mentioned above one has to accept the fact that if there is anything in test data, which was no present in training set, the model is bound to fail in that iteration of execution.

Finally when having all the check boxes verified, it can be thought to knock the door of machine learning.

The same is also available as medium post. https://medium.com/@deepak.neema/saying-no-to-machine-learning-cb5c8a17b662 I would be glad and eagerly waiting for your comments.