Don’t Fall For Data Science Myths, Says Anish Agarwal Of Royal Bank of Scotland

A data science professional with over 18 years of experience, Anish Agarwal, the director of data and analytics at Royal Bank of Scotland, India, is definitely one of the most influential leaders in the analytics industry. 

Machine Learning Developers Summit 2020

With a proven ability in designing and executing strategy and facilitating new technology adoption, Agarwal says, “A lot has changed over the last two decades. It almost feels like two centuries!” 

Analytics India Magazine caught up with Agarwal to get an insight into his analytics journey, some of the important contributions that he has made to the field, his advice for data science enthusiasts and more. 

Agarwal has been involved in various emerging innovations such as data analytics, strategy, financial modelling and artificial intelligence. He is also an avid speaker, has authored papers focusing on key AI technologies, and has been instrumental in implementing several new systems into organisations. 

As Agarwal shares, the term data science has emerged only recently to specifically designate a profession that is expected to make sense of the vast stores of big data. Having started his career in 2001, he shares that at that time data processing methods that were used tracked not only compiled data in rows and columns but also instilled curiosity amongst colleagues to understand the magic behind the bar and pie chart.

“Skill in MS Excel was most desired and anyone who knew Vlookup was adored no less than a superstar.”

Says Agarwal.

Agarwal confesses that he has always been good in the functional side of data and analytics as opposed to being an expert statistician with a good understanding of software architecture and multiple programming languages. Despite this, he boasts of a successful analytics career and definitely tons of things aspiring data scientists can learn from. 

Analytics India Magazine: With over two decades of a career in analytics, what have been some of the major highlights in your data science journey?

Anish Agarwal: While there are many, one of the most interesting high-points of my career in data science is what I call as self-discovery. This was when I was preparing for an internal job posting in one of the data science organisations I was working with, I came across two new words – data analysis and data mining which made me nervous! I felt having lost the game; technology seemed to be moving fast. But, after reading the definition, I took a sigh of relief as it was just a two-word definition of what I did in my previous role.

Today, we have simplified tools which allow decision-makers to become more self-sufficient. The tools are easier to use, provide the functionality needed, and are very efficient. Today, businesses can gather data and gain insights by working directly with the data.

Over the course of my career journey, I have developed deep expertise in defining a problem, identify the key sources of information, designing the framework for collecting and screening the needed data, and most importantly, clearing segregating the role of a computer and a human.

AIM: How did you land up a career in the data science industry? What are some of the tips you would like to give to data science enthusiasts who want to venture into this field from non-tech backgrounds?

AA: Well, my Data Science journey has been thrilling. The tryst with data starting with preparing sales report which got a push when I was tasked to add a few charts in my reports. It was in 2006 when I was asked to learn the data architecture of Siebel CRM. The organisation I worked for had decided to use this system to track all stages of the sales cycle. This is when I learnt the nuances related to terms like Business Intelligence and OLAP. Smart and jazzy, isn’t it? OLAP or Online Analytical Processing was a system that allowed analysing data from a variety of sources while offering multiple paradigms or perspectives.

And then, Business Intelligence went hi-tech and transformed into what we today know as a data science ecosystem comprising artificial intelligence, machine learning, Natural Language Processing, RPA and more.

Transition into data science can be tough, even scary! It is not because you need to learn Math, stats and coding but you also need to battle out the myths you hear from people around you and find your own path through them!

I had my own share of these “popular” myths, and they did make my life difficult. Some of these were;

• “You need a PhD to have a chance of becoming a data scientist. Two is even better!”
• “Participate in data science competitions, that will tell you how the industry works.”
• “You need tons of computational resources to build deep learning models. You can only get that at the top tech firms.”

Even today, there are multiple myths floating around that add a false aura around data science roles. Don’t fall for them!

These myths often make you feel like only geniuses can work in data science. This is just not true. Whether you’re a recent graduate, an experienced professional, or a leader, it’s important to understand how data science works and you will find your place in the industry. It’s important to understand the difference between the roles. In a vast and complex field like data science, practical experience is king. There are numerous projects you can pick up and work on right now. Or find a problem you are passionate about solving and see if data science techniques can be applied there. There are plenty of resources available online for learning.

Lastly, as a beginner, you will be required to learn concepts from scratch. It will often be a frustrating experience. Your technical colleagues might know more. They might be ahead of you initially at every turn. And that’s where the dedication and discipline characteristics we spoke about earlier come into play.

AIM: How important is it for data scientists to have business acumen along with technical skillsets?

AA: It is imperative that data scientists have a strong foundational knowledge, understand the business challenges, and how data and technology can be leveraged to bring out the insights.

As part of the data science journey, one will hear very valuable yet different perspectives on what it means to be a data science professional in the business setting, ranging anywhere from traditional data analytics and reporting to sophisticated machine learning model development and deployment.

Also, in pursuit of being precise in the world that isn’t, there are attempts to clearly state the importance of business knowledge. Your expertise and strength in both technical skills and business knowledge would only get you started. However, to make a mark as a data scientist, it is imperative that you develop the right business acumen needed for business growth.

Companies work on generating data touchpoints and analysing them for a reason – to understand crucial concerns associated with business operations and growth and recognise customer behaviour and deliver better services and experience. With the data you have in hand, you should be able to bring sense out of it and find angles and approaches towards business growth and potential.

What is important to understand here is the difference between knowledge and expertise. You don’t need to be an expert in the business, but knowledge is important.

AIM: What are some of the key challenges of setting up a data science team in organisations?

AA: Challenges are not limited to data science but are common in all areas of the business. Data science today is applicable throughout every industry. I no longer must explain what I do for a living as long as I call it AI — we are peak data science hype!

To make use of big data, businesses have started to either amass or hire teams tasked with deriving insights from the plethora of data collected each day. The demand for that archetypical “Data Scientist” who is the perfect blend of a statistician, programmer and communicator have never been greater. But as the dust settles, we have started hearing stories of failed projects and disenchanted professionals.

Many of the challenges are centred on the need for collaboration between technology and business. Common mistakes, such as using static data or not thoroughly planning a solution’s implementation, can trip up a young data team before they complete their first proof-of-concept. As data teams mature, the challenges do not go away but instead take different forms, like deciding whether to stick with older technologies (SAS, SPSS) or opt for newer approaches (R, Python, Spark).

Another challenge faced after setting the team is the unconscious bias around working with blinders on, focusing entirely on meeting project deadlines and not realising that the operating premise itself has become flawed; the foundation is not stable. Dynamic data rears its ugly head as changes in technology, languages, and stacks invalidate the usability of time-sensitive solutions.

But when these teams are well organised and given the right tools and direction to succeed, they can do much more: data teams can serve as a research and development function, experimenting with raw data to explore possibilities and solutions that the business didn’t even know it had. This can include anything from uncovering new insights about customer behaviour to revealing business opportunities in new markets. 

AIM: How is the AI startup community shaping up in India? Is there more hype than actual usage of AI by these startups?

AA: Startups, in India and in many other parts of the world, have received significant attention in recent years. Their numbers are on the rise and they are now being widely recognised as important engines for growth and jobs generation. Through innovation and scalable technology, startups have an opportunity to generate impactful solutions, and thereby act as vehicles for socio-economic development and transformation.

Today’s AI start-ups do their best to emulate the functioning of the human brain’s neural networks, but they do this in a very limited way. To put it simply, they teach the algorithm what they want to learn and provide it with clearly labelled examples, and it analyses the patterns in those data and stores them for future application. The accuracy of its patterns depends on data, so the more examples you give it, the more useful it becomes.

Herein lies a problem: An AI is only as good as the data it receives, and it can interpret that data only within the narrow confines of the supplied context. It doesn’t “understand” what it has analysed, so it is unable to apply its analysis to scenarios in other contexts. And it can’t distinguish causation from correlation. AI is more like an Excel spreadsheet on steroids than a thinker.

Part of this deception is about riding the wave of pervasive media attention AI receives, a typical corporate eagerness not to miss the train. “It’s a buzzword”. Every day there is a bunch of news about AI, like AI beating humans in doing this and that. It’s just easy to jump into that trend.

Most of the AI startups being founded and funded last year in India were into AI-enabled applications for multiple use cases in different verticals. Now, if these applications can go deep enough, thanks to top-notch Indian tech talent, we could see a tipping point for AI startups from India this year or the next.

For Indian start-ups to take maximum advantage of AI and ML, large scale actionable data needs to be available to train algorithms. While the ongoing digital transformation of public and private sector enterprises transform the economy from being data-poor to data-rich, India’s strong talent ecosystem would help us accelerate the adoption.

AIM: What are some of the key requirements you would like to suggest for the founders of AI start-ups to have?

AA: Founding a startup can be hard. From financing your endeavour to having a skeleton staff and navigating the legalities that come with starting a new venture. Artificial intelligence is, of course, all the rage in tech circles, and the press is awash in tales of AI entrepreneurs striking it rich after being acquired by one of the giants, often early in the life of their startups.

The path to success in AI requires not just technical prowess but also careful thinking and execution through a range of strategic and tactical questions that are specific to this domain and market.

Here are a few tips:

Do your research – Today, we’re in a much less forgiving startup ecosystem. If your competitor hits the market with a product, they can get it in front of hundreds of thousands of people via the internet and social media in a day. Whatever your idea, you can be sure there are a few other folks working on it, even more thinking about it, and thousands who will jump in as soon as they hear about it. Before you burn through your life-savings funding a startup, you first need to find out if there’s a market for your idea and what its chances are of being successful. According to the experts, one of the biggest mistakes new founders can make is misunderstanding the size of or other details about the market for their idea.

Understand how to get funding – There are several ways a fledgeling business can help raise some much-needed capital. Some startups with an idea for a popular consumer product may seek out funding through crowdsourcing, whilst others may seek angel investing. Some, more mature startups, may instead seek funding from venture capitalists. It is important to keep in mind however that funding from the latter is usually tied to stringent expectations about high rates of growth and returns.

Are You Sure You’re Building an AI Startup? – This is one of the more difficult or annoying questions to answer (depends who’s asking it and of whom). A better form of the question might be: “Can your startup be done without AI?” or, even more precisely, “Can your startup be done without building the AI components yourselves?” Even if your company’s vision is to use AI to deliver value, you need to ask yourself whether the AI needs to be proprietary. If AI is central to what you do, then there are good reasons (e.g., accuracy, long-term costs, data privacy, tackling new challenges) to consider building things in-house.

And most importantly, are you ready for the startup lifestyle? If you are currently an employee of another company, then starting your own as an entrepreneur is a lifestyle change. Don’t make the mistake of assuming it is a way to get rich quick or an escape from all the problems. Starting a business is hard work, requires a lot of determination and learning, and only pays off in the long term. Take an honest look at yourself before leaping.

AIM: Is the banking industry witnessing analytics and data science adoption to the extent that it should? What are some of the areas where it can be adopted?

AA: In today’s competitive world, growing a customer base and satisfying them is considered the most challenging task. Customers demand are being treated as individuals and not as a general lot. To get over this, banks have been implementing various tools over time.

Challenges like ensuring long-term loyalty from high-valued customers, retaining and attracting different types of customers or cross-selling of which products exactly to whom, fraud detection, application screening, credit and collections has always been an area of concern. Analytics comes into the picture here. It helps banks to fetch the relevant data of customers, identify fraudulent activities, helps in application screening, capture relationships between predicted and explanatory variables from past happenings and uses it to predict future outcomes.

The banking industry is data-intensive with typically massive warehouses of untouched data. New models of proactive risk management are being increasingly adopted by banks. By using analytics to extract actionable intelligent insights and quantifiable predictions, banks have learnt to gain insights that encompass all types of customer behaviour, including channel transactions, account opening and closing, default, fraud and customer departure.

While analytics isn’t exactly new to the world of banking, plenty of banks are gearing up for their next big analytics push, propelled by a load of data and new, sophisticated tools and technologies. The key business drivers which increase the importance of analytics within the banking industry are Regulatory reforms, Customer profitability and operational efficiency.

Banks are also harnessing the power of analytics in credit scoring applications to more accurately estimate the risk associated with a potential customer. Most credit scoring methods consider the potential customer’s credit and financial history. The architecture behind the credit scoring based lending model allows banks to base their credit scoring on alternative data types such as social media posts and interactivity. This could include what sites a potential customer visits, what they purchase via eCommerce, and what they say about those sites and purchases on social media. The online behaviour of a potential customer can indicate the likelihood that they will pay back their loans and make payments on time. The sentiment becomes a data point indicating a “positive” or “negative” experience, which can then be recognised by a predictive analytics application.

AIM: How do you see the analytics industry evolve in the next five years?

AA: This is a terrific question with a nuanced answer. If, by analytics, we mean the specific process of taking data and explaining what happened in it – and strictly following the theory which describes the trajectory of technological evolution, then analytics as a profession will go away in five years.

If by analytics, we mean the general process of taking data and extracting actionable insights, then machines will do the analysis, we’ll provide the insights, and machines will execute on our findings. Machines are becoming more and more capable of explaining what happened. Data analysis software that cost millions of dollars a decade ago is open source and free to anyone skilled enough to implement them today.

Tools like Watson Analytics have the power to make detailed analysis inexpensive and accessible to anyone. NLP will generate literature reviews with no misses. Our systems will be more than capable of creating anything we need, be it automation, insights or visualisation.

In the hierarchy of analytics, we are entering the predictive era. In the next 5 years, I expect us to fully embrace predictive analytics and begin venturing into prescriptive analytics.

What’s definite is that data & analytics will only gain momentum for the foreseeable future and will be at the core of countless new technology solutions. Analytics will continue to focus on usability and increasing natural language that enables business users to extract data and generate insights without needing to understand the underlying algorithms. Not only will this increase efficiencies and create further adoption throughout companies, but it will also help alleviate some of the problems created by the data scientist shortage.

Furthermore, the opportunities created by machine learning and artificial intelligence are endless, and it will be a race for companies to harness its power and create new services that provide value in unique ways.