Now Reading
Data Science in Developing Economies: A Travesty of Business Investments? – Part 2 of 2

Data Science in Developing Economies: A Travesty of Business Investments? – Part 2 of 2

Mumbai, India - January 5, 2015: Crowd At Chhatrapati Shivaji In

Read the first part of article, here.

So what should Analytical systems look like in emerging economies with nascent data infrastructure?

We will start this section with a conjecture on what may be the pitfalls of a purely subjective / heuristic based decision process in emerging markets. Overtly the risks involved are minimal – so long as a breed of very capable and effective decision makers are available to take the right decisions. However, this situation by its very nature appears to be serendipitous. There is obviously no guarantee that subjective capability is sustainable in organizations in the long run. Hence, some elements of objectivity in the system in terms of a process or, a plan to build one in the long term is required to act as an insurance against possible scarcity of high calibre leadership in the organization.

The presumption here is that data science related processes, in spite of their limited capabilities due to the constraints cited earlier do have a role to play, either in terms of partial support, or validation of hunches, or in the best case scenario a new revelation that may have been hidden by normal subjectivity.

However, before we delve into the virtues of data science further, even in the realm of sparse data availability, let me point to some possible action steps for organizations in India (developing world) to consider, especially when there are no clear directions for building appropriate analytic ecosystems.


We list out some guidelines that may be useful for analytics practitioners in many Indian organizations. These are largely based on our experience of working with organizations wanting to build analytical decision support systems, but are constrained by having unorganized and distinctly incomplete information resources.

1) Ignore the data scientists for the time being but, imbibe their approach

Many doctoral students in management (including the author) train themselves to become rigorous management scientist – they work hard on mathematical problems with fleeting relevance to the management world and devised seemingly eclectic solutions for very narrowly defined problems that perhaps had very little to provide in terms of assistance in real life applications. However, in spite of the many naysayers of this kind of training, it does have an impact in creating a knowledge base which over time trickles down to the application domain. Additionally, and the point that needs emphasis here, it provides a very strong foundation for critical thinking and the clarity in identifying the most appropriate dimensions for problem solving. Data science (and more generally, decision science) has a clear approach towards identifying relevant dimensions, their associations and subsequent impact on business performance. The depth in the approach helps provide perceptive thinking ability, which is essential for developing an effective analytical-driven process for decision making in the long run. Hence, data scientists are required in organizations not just for their business model building skills, but to apply their generic logical problem solving skills towards building better information supported decision making process in the long run.

2) Develop a medium range strategic analytics plan for addressing topical business questions

The idea here is to use the data scientist’s skills to develop medium range plans to address ongoing business challenges with factual information. To initiate, a) identify what may be very desirable information and seek out potential sources of such information and, b) Identify potential analysis at a high level that may be applied to get answers to business questions that require inputs. It is important that such exercise be done comprehensively and the actual paucity of data needed not be used as basis for making a sub optimal plan. The time for making compromises in implementation will come at a later stage.

Organizations must develop a plan for the ideal situation which acts as a template for identifying ongoing opportunities to improve.

3) Identify near term opportunities for supporting decision making with tactical and “piece meal” analysis

All planning exercises require a “fall back” option to ensure that expectations are met realistically. Obviously, we should not dilute the importance of comprehensive planning – since they are a roadmap for future enhancements. However, realistic expectations in the near term are set based on availability of data and more importantly, the nature of insights that they provide (under specific assumptions) to help in decision making. This planning activity is very necessary to claim organizational support for initiating near term investment in analytical warehouse.

See Also
IBM 2020 plan
IBM To Double Down On Data, AI and Cloud Services In 2020

4) Apply your organization’s data science resources judiciously to provide factual support for decision making as identified in the previous step

We are perhaps over the hill in terms of organizational acceptance of analytics and its role in decision making. The rest is to make it happen and show results and benefits of the process. The motto is to start simple and here the skills of data science are very helpful. More importantly, data scientists are expected to parsimoniously (but effectively) provide insights that are helpful for decision making with necessary caveats for the decision maker to appreciate the true value of the information. Hence, the role of the data scientists is not just to churn the data, but also provide meaningful inferences in the context of the decision that needs to be made. This may also involve applying qualitative assessment and intuition based on past experience of analysis and also conducting experimentation to confirm partially developed hunches. This capability of building business oriented inferences looking a network of information across many and disjointed sources (both factual and biased), is a capability that is often times overlooked in today’s technology-led forays in analytics.

5) Look for opportunities to enhance data warehouse with newer sources of internal and external data that can support better analytics infrastructure

Last but not the least, for long term sustenance, the process of enhancing business data warehouse is an ongoing one. As markets develop and collection infrastructure improves, the organization must step in to leverage the advantage of newer and better information and augment its current analytical processes. In the process, ensure the fulfilment of a more comprehensive decision support system. Hopefully, that would be the harbinger to enhance the role of data science in business decision making in developing economies.

Concluding Remarks:

This is an exacting set of actions for most organizations. However, with increasing competition and lesser market space to manoeuvre, the sooner organizations tread down this path the better for them to retain their precision in decision making. Surely, the implementation of data science in a direct sense is easier, provided the data are available in a form that readily lends themselves to processing and insight generation. For most organizations in developing economies, there is much work to be done in data management preceding such direct and impactful use of data tools. Barring some domains such as digital commerce (which are about less than 1% of the total transaction in India3), most business organizations do not have the luxury to an organized access to information. Hence, Analytics processes in countries like India require a different focus compared to what is seen in the developed world.



Provide your comments below


View Comments (0)

Leave a Reply

Copyright Analytics India Magazine Pvt Ltd

Scroll To Top