CIOInsider India Magazine
Data Science & Analytics - The Indian Scenario
The global data science market size is expected to reach $140.9 billion by 2024. How do you see this market growing in India? What will be the major factors driving this growth?
India, like the rest of the world, is witnessing a rapid growth in the use of Data Science to drive superior business results. There are few underlying reasons for this trend
•Increasing Digitisation of our world is creating new data streams, which are now being captured, stored and mined for extracting valuable business insights.
•As the benefits of Data Science become more obvious, more and more companies are adopting the use of Data and Analytics in their everyday work.
•The breadth of Data Science applications is also growing rapidly, as newer and newer methods such as IoT analytics become more widespread.
•Most companies have progressed from ‘proof of concept’ projects to serious engagements either by setting their own Data Science functions or by partnering with Data Science providers
To sum up, this increasing appetite for Data Science in India, which is further growing and is likely to continue into the future.
Looking back at shoppers’ past activity often isn’t a good enough indication of what they will do next. Historical data analysis often delivers imperfect analysis and outcome. How should data analytics be processed, particularly keeping in mind the diversity and variables of the retail industry?
Historical data, if analysed correctly, can indeed make useful predictions about customer behaviours. However, in order to do so, we need to leverage the combination of having the right data, deep domain knowledge and the technical expertise to use Machine Learning algorithms. In order to ensure that the right data for all use cases is captured, processed and stored for easy access, a comprehensive Data management strategy needs to be adopted. Without creating this foundational element, no comprehensive Data analysis is possible.
Next, for complex use cases, it is important to adopt a ‘man in the loop’ approach, where a domain
expert’s knowledge is used to validate hypothesis, understand causality, select modelling variables, and more. This approach ensures that the modelling and analysis truly reflects real world scenarios and therefore results are accurate as well as consistent.
Finally,today’s advanced Machine Learning algorithms do have the ability to learn ‘weak signals’, interpret complex patterns and leverage large data sets to make accurate predictions. However, we still need a capable team equipped with the right tools and platforms to utilize these technologies effectively.
Historical data, if analysed correctly, can indeed make useful predictions about customer behaviours
Supply chains are already driven by numbers and analytics, but retailers have been slow to embrace the power of real time analytics and harnessing huge, unstructured data sets. How should the SMEs approach to leverage data science in their operation?
For evolved firms, who are already leveraging their supply chain and production data for insights, exploring real time and unstructured data analytics can certainly be very useful. However, it is important to put the horse before the cart. We need to adopt a business first, not a technology first approach. In other words, it would be useful for SMEs/retailers to first map their current supply chain challenges or opportunities to potential analytics solutions to determine the need for real time or big data use cases.
If in this process, the need for real time analytics or big data arises, then it becomes worthwhile for retailers/SMEs to invest and adopt such technologies. In fact, leveraging these methodologies for evolved companies can have long term transformative effects, if done correctly.
With the scale and complexity of modern data, the only way to truly harness the value is to automate the process of data discovery, preparation and blending of disparate data. How can this be achieved?
In my view, it is difficult, if not impossible to completely automate the entire data discovery process. Even with sophisticated AI/ML algorithms, it is generally accepted that domain knowledge and the Data Scientist’s experience plays a crucial role in delivering more real world usable and consistent solutions. However, in the more operational aspects of the Data Scientist’s role such as preparing data sets, or testing and scoring models, automation is playing a key role. Today, automation in the processing,storage and serving of data or autonomous model life cycle management for managing ML algorithms is helping Data Scientists work more effectively.
