Importance of Analytics in Today’s World
Analytics has grown rapidly in recent years due to the proliferation of digital data and technological advancements. As data becomes more important in our lives, analytics is expected to become increasingly important in businesses. According to MicroStrategy’s Global State of Enterprise Analytics Study 2021, 93% of organizations use analytics to influence business decisions. Another survey by Deloitte discovered that businesses that use analytics are twice as likely to be in the industry’s top quartile of financial performance. Analytics has several business ramifications, including better decision-making, higher competitiveness, improved customer experience, cost savings, risk management, and new revenue streams. The relevance of analytics in driving corporate success is increasing daily as firms collect and analyze data. To throw more light on the importance of Analytics for business, we spoke with Prof. Venkataraghavan K, Associate Professor, Information Technology and Systems Area, IIM Kashipur. A podcast version of this interaction can be heard on our Spotify channel, Vimarsh.
I:
The scope of analytics is vast, and organizations can use big data analytics systems and software to make data-driven decisions that can improve their business-related outcomes. The benefits may include more effective marketing, new revenue opportunities, and improved operational efficiency, so if I ask you, owing to the recent developments in the domain of analytics and big data, how do you see the market changing, and what are the users looking for?
VK:
Big data has been around for a decade, but in the last five years, organizations have increasingly adopted it. First, I will answer why we need to tame big data; organizations run several internal processes and external ones. While running these business processes, organizations do generate many data. When we say, “many data,” they come in various forms, they come in various sizes, and they come in various velocities. Now, you need to have an ecosystem to collate this data first. Only if you can collate this data, then can you run some analytical models or some queries on this data. Then get certain insights from this data. So, how do we do it? We do it by using big data technologies. Big data technologies help in two ways. Firstly, they provided a distributed computing environment, and secondly, they provided distributed storage environment. When we say distributed computing environment, you now have the power of thousands of commodity computers to process your data. Likewise, when we say distributed storage, you now have thousands of computers or storage devices to store your data which would not have been possible in a single computer. Big data is helping our organizations to tame the data and then extract insights from data. So how do they do it? Organizations benefit from big data in three different ways. Firstly, at a broad level, it could be business model innovation. Second, it could be process innovation. Third, it could be product or service innovation.
I:
Well said. This is now self-explanatory; how big data is playing a very vital role in our lives, and organizations do need to know how they can leverage more out of it. As you have said, on the same note, I would like to ask how you would recommend companies should approach analytics and get started with it.
VK
When considering adopting analytics, technology will not be a constraint; that will be the least of an organization’s problems. So where exactly does the problem lie? It is a threefold problem. One: there is an inertia to change within an organization. Organizational change is the single factor restraining the adoption of innovation. The second problem, we can say, is knowledge gaps among the leadership in the organization. The third one is the uncertainty of the business value the organization can acquire by adopting analytics. So, these three points are not necessarily what you can say are technology-related but have something to do with change or some gaps in knowledge people in the organization have. How do you overcome these problems? The first is experimentation, wherein the organizations can take in small projects and then see the business value of the small projects. Once there is sufficient value in going ahead, they can go for full-scale analytics deployment or implementation of technologies that fetch the value from the data using certain tools and techniques. Second thing, how do we overcome the knowledge barrier in the organization? That can be done through sensitization programs in the organization, so senior leadership or people who are in the driving seat in the organizations can look at programs and they can understand how technology can help them. That is one approach wherein they can overcome the knowledge barrier. Then they can also look at hiring experts. Organizations have people who have been working in them for many years, and they have a good amount of domain knowledge. But, they lack the technical expertise to tap into that domain knowledge to develop useful innovation. How can you overcome that? You can overcome that by hiring experts with profound knowledge of that technology. One more thing an organization can do is create cross-functional teams in the organization. Cross-functional teams put like-minded people together who can work on a particular project and then get some useful outcomes from the project. For example, you can bring in data scientists, you can bring in data engineers, you can bring in BA experts together in a team and then work on particular business problems and then solve it and then come up with certain useful outcomes for the organization. So, build cross-functional teams, remove your knowledge gaps, hire knowledge experts, experimentation. These are some ways an organization exploring adopting analytics can implement its initiatives.
I:
So, as you mentioned, technology can be learned, but nowadays, organizations are more focused on Tech experts, and they are investing in senior leadership towards technology. So, what is your thought process on that? Whether they should be looking toward the business evaluation side or towards the technical aspects.
VK:
See, it’s a mix of both, there are certainly good practices that organizations can follow in adopting these analytical tools and technologies and then going ahead with any analytics project. First, do not worry about technology. Organizations must first look at the business value of any initiative. If they are sufficiently satisfied that there is some business value, then they can embark on a project, and there are certain industry standard practices that have been followed for almost two decades now. One good example is CRISP-DM, which is a Cross-Industry Standard practice for Data Mining. It gives a sequence of steps any organization can follow. Starting with business understanding, then getting into data understanding, then model development, model evaluation, and finally deployment. So, organizations first need to worry about a business problem. Get the scope of the business problem clearly and get the requirements. Once the scope and requirements are clear, you determine the data required to solve that problem. And when looking at the data, you must look at many things. Is the data of good quality, or are there certain governance practices as far as the Data is concerned? You need to understand these things. Then adopt appropriate tools. Appropriate tools, meaning algorithms and techniques, can use that data to answer that particular business problem. In this case, you follow the entire process of machine learning, deep learning, or any other related process. You can build models and then test the models, and when you are sufficiently sure that yes, the models which you have built are good, then take the feedback of the business and then go ahead with deployment.
I:
So, since you mentioned artificial intelligence and machine learning, my next question is when it comes to experimenting with AI and ML across the three different adoption stages, the experiment, the move, and improving and building analytical models from scratch. Are there any best practices that you can highlight? For each of these three different journeys regarding AI and ML.
VK:
As I just answered in the previous question, one of the best practices is CRISP-DM practice, which gives a sequence of steps. You go through the entire journey from business understanding to deployment. One more thing I would like to add to what I had said previously is that even after deployment, there is no guarantee that the models you have built or the solutions you have developed will keep working fine. There is no guarantee that even after deployment, the models we have deployed will continue working fine. That is because the data will always drift. There will be changes in the data because the data is dynamic in nature. So, after deployment, we need to monitor our data sources continuously. Suppose there is any variation in the model’s performance. In that case, you repeat the entire cycle, keep fine-tuning the models or building new models, and ensure that the models can work with the changing nature of the data or accommodate the dynamic nature of the data.
I:
Since we have already discussed analytics, we will also discuss the job industry in an analytics role. So, there are many ways to start a career in analytics. Which one do you think is best, or which way is best for individuals to fit into the current AI-ML job market? What would be the best strategy and tools one should be comfortable with?
VK:
That is an interesting question. Looking at the job market, especially the AI-ML market. The first thing is not to worry about the tools. First, get a good understanding of the fundamentals. If you are planning to become a data scientist, you need to have a good understanding of the basic sciences like statistics, probability, applied mathematics, machine learning, and deep learning disciplines. Once you have a good grasp of this, you worry about what kind of tools or technology you can use to build models. The first thing is that you get a strong hold on the disciplines. Once you have a strong hold on the disciple, you specialize. One cannot become an expert in all the domains and verticals. You must choose a particular domain or a vertical and dive deeply into it. The second part is the domain. You have to choose the domain in which you want to work carefully. Another aspect is vertical. There are thousands of data science techniques. We will not be able to specialize in everything, so you need to choose a vertical for yourself carefully. If you are passionate about images or computer vision, work on computer vision or image analytics. If you are passionate about text mining, then work on text analytics. Or if you do not want to work on images or videos or, say, text if you are interested in working only with structured data, you can specialize in certain tools and techniques which work with structured data, so you must carefully plan your journey starting from the foundation, then choosing your domain and then choosing a vertical. Tools are not something you need to worry about; it could be Python. It could be R. It could be some other tool. Learn the concepts and get a strong hold on these concepts. You can then choose a tool of your choice and then implement these things.
I:
There are several resources available on the Internet also, and one can move in that direction. When discussing analytics, we always try to associate it with certain industries. Taking the example of the Ed-tech industry, we have seen a recent trend where Ed-tech takes the concept of traditional education and gamifies the learning processes using the latest technologies such as artificial intelligence, machine learning, and others. So, what do you think about the AI-powered EdTech sector?
VK:
The EdTech sector, in general, has democratized education or learning in a big way. These things may not have existed maybe a decade earlier. But today, you see hundreds of opportunities for learning, and in them, gamification is a novel concept as far as learning is concerned. But gamification has been there for many years. Before the advent of edit texts, gamification was employed in organizations, it’s a way to keep one interest going. It is easy to start something and lose interest when considering learning, so gamification helps. Gamification is one way to retain the learner’s interest.
I:
And yes, that’s why AI ensures multiple checkpoints to these industries so that they can evaluate the content and excellent recommendations. So, as we have talked about industry analytics, now we will come to IIM Kashipur. It also started the MBA and Business Analytics program in the year 2020 specifically, which is designed for students fraternity who wish to join the big data revolution and emerge as future leaders in the field of data analytics. So how is this program helpful in this holistic development of students who strive to become highly efficient managers with an analytical mindset and an inclination towards data-driven decision-making?
VK:
Let me start the answer with the background of MBA Analytics at IIM Kashipur. We started the MBA Analytics program in 2020, but the analytics history goes back to 2016. We have been running an analytics specialization or analytics stream, we used to call it ‘analytics stream’ earlier, and we used to expose students to analytics courses that were most sought after in the industry. In that way, we also learned that IIM Kashipur has been equipping itself in this area for many years…. from 2016 onwards. And in 2020, this culminated in the creation of a new program name ‘MBA Analytics. How is MBA Analytics distinct from the regular MBA offered at IIM Kashipur? As the name suggests, it’s a program focusing on the analytics discipline. And how do we prepare the students? In the first year, we have two tracks. One is a management track, and the other is an analytics track. The students get to do management courses and analytics courses as well. So that’s a nice blend that is offered in the first year. When you are looking at the analytics track, there are several courses or subjects which provide a good foundation for the students, right from statistics, mathematics, programming languages, and applications of analytics. So, students get prepared for their journey in their second year. And in the second year, they tend to take up specialized electives, which cover the domain and the vertical. So, I spoke about the domain and vertical a while back. When we talk about domains, there are many electives, such as sports analytics or health analytics, which are offered as electives. Likewise, when looking at the vertical, there are many things you focus on, say social media; you have analytics courses specializing in structured data, natural languages, computer vision, and image analytics. And then, there are marketing, finance, supply chain, and HR analytics. So, the second year is arranged in such a way that it gives a nice blend of both domains and verticals. In that way, the program prepares the students for a long career in the analytics industry.
I:
Thank you, Professor, for giving us the opportunity to interact with you and learn much helpful information.
(To listen to an unabridged version of this podcast, go to IIM Kashipur’s official Spotify page ‘Vimarsh’ or click here.)
