We are living in the age of Analytics. The buzzword surrounding us are artificial intelligence, machine learning, deep learning, computer vision, image processing etc. There is an overwhelming enthusiasm around these buzzwords across all industries like Banking, Insurance, Healthcare, Telecom, Retail, Manufacturing to name a few. Some of the industry verticals have been using analytics for a long time and the other ones are gradually getting aligned to emerging tech. There is more focus on this domain than ever before.
This leads us to a very critical question. Do we need an AI solution to all our business problems? All industry leaders across verticals are focused heavily on AI and advance solutions involving Deep Learning and Neural Networks.
Subscribe to our Newsletter
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
While there are critical and complex business problems which may require the involvement of AI and Deep Learning, the success of a data science project depends upon simple building blocks of logical thinking and mapping the logic to the business problem.
For example, if we are trying to predict whether a Promoter or Builder should build a housing project at a certain area. The parameters to think will be Electricity availability, Water availability, and other basic amenities in and around that area. Suppose an area is projected to have an acute water problem in the next three years. Solving this problem will require a simple decision rule-based system. We may not even require a statistical model to fit in to solve this problem.
Let us take another instance, where our business problem is to extract relevant information from scanned images that contain handwritten text. There are multiple complexities involved in this problem. We could break it down into 3 parts. The first part is reading an image. The second one is identifying the part where there is a handwritten text. The third one is to extract the exact text through optical character recognition. After these three steps are achieved, we will have to apply our own logic to extract the required information. This is a problem which is much more complex than the earlier one. Advanced algorithm implementation can be a good fit to solve this kind of a problem.
What Does A Data Scientist Do?
A Data Scientist primary work is to understand data and learn complex algorithms to keep up to date with the advancement in technology. But the responsibility does not end here. He has to understand the business context and apply an algorithm that is best suited to solve the business problem. The algorithm can be very basic or it can be an advanced AI-based system.
Another aspect to keep in mind is the infrastructure cost. In recent years the cost of infrastructure has decreased. As such more and more businesses can afford to build AI solutions. Ideally, a Data Science project should have an ROI (Return on Investment) calculation before starting up. Building a simplistic solution with negligible infrastructure cost can sometimes give us a better return than a complex solution. Also, maintenance of a complex solution in production may involve some cost along with it. A complex neural network or deep learning algorithm always comes with a cost.
The Biz View
From a business point of view, it is always better to think with a clear head. It is always wise not to get overwhelmed by industry buzzwords. Every solution to a business problem starts with simple logic building exercise. Remember, Data Science is a science and there is no magic in it. Every algorithm be it simple or complex is driven by mathematics.
There is an overwhelming enthusiasm among practitioners in India to switch to Data Science. A bit of motivation comes from the fact that Data Scientist gets paid better than practitioners working in another domain. However, as this is a multidisciplinary subject acquiring the skills to be a successful data scientist requires time, a lot of hard work and dedication. Programming skills, Statistical knowledge and some consultative skills are the building blocks of a data science professional.