The Art of Analytics

We are in conversation with master data story teller: Anand, who is the co-founder of Gramener, a data science company. He leads a team that tells visual stories from data. He is recognized as one of India’s top 10 scientists, and is a regular TEDx speaker. Anand is a gold medallist from IIM Bangalore and an alumnus of IIT Madras, London Business School, IBM, Infosys Consulting, Lehman Brothers, and BCG. More importantly, he has hand-transcribed every Calvin & Hobbes strip ever, is addicted to Minecraft (thanks to his daughter), and dreams of watching every film on the IMDb Top 250 (except The Shining). He blogs at s-anand.net. His talks are at bit.ly/anandtalks.   Highlights: [00:00:36] When moving from London to Bangalore, my co-founders were nudging me to move to hyderabad. So I did what I ought to do, which is pull all the data from GitHub, scraped the developer dataset and looked at the number of developers in Bangalore.[00:02:00] Every year there's 60% more data that's generated than the previous year.But if you look at the growth of the analytics industry, that's only about 30%, which means roughly half the new data generated every year not analyzed.[00:04:08] but what are stories? Simple- Insights connected together in a sequence. [00:13:16] The best dashboard that I've seen is the non-existent one. For one of our clients, each of their sales team gets an email. They don't have to see the dashboard. They get an email that says, here are the three contracts that are going to expire the coming month.[00:19:47]  You can construct charts out of smaller elements, just like you can construct a dashboard out of charts. [00:27:23] I can always define a standard of data that cannot possibly be met. Because the real world is messy.[00:27:40] Question is, does that number even define what I want it to mean? Let's continue the example of delivery time. What constitutes delivery time? The time at which it was left at the doorstep, or the time at which the customer picked it up?[00:32:13] Now if we can't stop lying to ourselves, lying to others is just so much easier. [00:35:00] One of the things that we find fascinating is comic based data storytelling.  Resources: Team of RivalsThe Grammar of GraphicsComicgenSlideSenseExplorable NarrativesUrban Heat Islands of CalgaryThe Data Detective

Calvin Chu Yee Ming is Managing Partner at Eden Strategy Institute, LLP. He has advised senior executives in over 20 countries, including organizations such as 3M; Bell Labs; Canon; Coca-Cola; Cummins; DBS; DHL; Disney; Fujitsu; HP; Intel; General Electric; M1; MasterCard; Medtronic; Nikkei; Nokia; Reed Elsevier; Roche; Samsung; SKF; StarHub; Standard Chartered Bank; Sumitomo; TNT; UNDP; UNESCO; UNICEF; and VISA; as well as the governments of Australia; Malaysia; New Zealand; Indonesia; Philippines; Singapore; Thailand; and the Kingdom of Saudi Arabia. Calvin was recognized as a NetImpact Change-maker in 2014 and inducted into the International Who’s Who of Professionals in 2009. Under Calvin’s leadership, Eden Strategy Institute has been awarded as The Most Innovative Management Consultancy in APAC Insider’s Singapore Business Awards; Corporate Livewire’s Management Consultancy of the Year; Global 100’s Most Innovative Management Consultancy (Singapore); Corporate Vision’s Best Social Innovation Consultancy (Singapore); and was also the winner of the National Business Award in the Consulting category by the Singapore Business Review. His work has appeared in Asian Banking & Finance, Asia Pacific Biotech News, BusinessWeek, The Star, the Straits Times, the Singapore Business Review, Today, and the Wall Street Journal, and he has featured at the ASEAN Smart Cities Network, ASEAN Social Entrepreneurship Forum, The Economist Social Innovation in Action, the Regional CEO & CIO Summit, Asia-Pacific CFO Summit, Business & Nature Forum, Institutional Investors APAC Summit, Singapore Business Federation, TEDx, Education Innovation, Prepaid Mobile Asia, Private Healthcare Asia, and Biomedical Business Conference. Calvin has been a Judge, Reviewer, and Mentor at the President's Challenge Social Enterprise Award, The Grand Challenges Explorations (GCE) Program of the Bill & Melinda Gates Foundation, MIT Inclusive Innovation Challenge, MIT Emerging Technologies Innovators Under 35, the Youth Social Enterprise Programme Grant Committee, The DBS-NUS Social Venture Challenge Asia, The Grameen Creative Lab, the Lee Kuan Yew Global Business Plan Competition, Social Innovation Camp Asia, Start-up@Singapore, and The University of Chicago Booth School of Business Global New Venture Challenge. He has also served as an iAdvisor with IE Singapore, an Executive Advisor at NUS Enterprise, and on the boards of BioFourmis, Bettr Lives, Conjunct Consulting, Rotary Club, and the World Toilet Organization.   Highlights [00:01:00] Profit of course is important. It is critical to be able to drive any kind of outcomes that we're looking for. But at the same time, we also want to have a line of sight to say that, you know, if we are working on this piece of work, for example, in healthcare or in education or in smart cities there is a prospect of creating shared economic and societal outcomes. So that's what we call social innovation.[00:09:00] Digital trust is basically the confidence that we can give to stakeholders, that you can do business with each other, whether consumers or citizens or fellow businesses, in a trusted, frictionless, dependable manner.[00:11:00] What would be the right governance structures for a company holding data when its supply chain is parking the data in many other jurisdictions under the regimes of other governments?[00:14:00] A big bank and a big telecom company, both sitting on a lot of data, if they could only come together, it is magic, right?[00:17:00] Trust in organizations was never lower than it is right now. And this is a function of geopolitics… And that's on a global scale, right? So, there's definitely a need to do a lot more of this.[00:21:00] I could be Chief Marketing Officer, Chief Operations Officer, or Chief Technology Officer and I should interpret the sustainability mandate within my own job description. That is the only way that companies, these elephants, can dance, right?[00:26:00] We need to think of a system. How do we have a circular strategy? … So again, getting data, analysing, optimizing these things… these are very vital to help people see these linkages rather than to take a very blunt, a mental shortcut, right? If I have EVs is good, if I go vegetarian is always good… Everything has its repercussions.[00:30:00] We take a more humanistic approach to really understanding why are people not moving and how is it that we can actually motivate them internally to want to move.[00:36:00] And while it can be tempting to think of (social) concerns as if they were costs, when you can angle them and flip them and reframe them the right way, they all end up being assets for you to make altogether better strategy.   Resources Eden Strategy Institute: https://www.edenstrategyinstitute.com/Digital Trust white paper and context: https://globalfutureseries.com/digitrust/wp/download-white-paper/https://techwireasia.com/2021/09/heres-why-sgtech-set-up-a-digital-trust-committee/  Edelman Trust Barometer: https://www.edelman.com/trust/2022-trust-barometer

They call themselves a process safety company but there is more to Empirisys than just safety. Listen to industry veteran Pete Sueref talk about how data analytics and behavioral science can combine to help prevent the greatest man-made disasters of modern times. Wouldn't you like to know what really caused the challenger disaster?   Highlights [00:03:30] human factors or organizational factors ... are the root cause of pretty much every single huge catastrophe. [00:05:21] there's quite a harmful, safety artifact that you see in lots of buildings... we haven't had an accident here for a hundred days, and you update her every day and 101 days... [00:06:34] there's a whole movement in the industry around safety one versus safety two, where safety one is looking at fault and blame and safety two ... improving and taking ownership for safety in the organization.[00:13:00] as well as being a data science consultancy, we also have leadership and behaviors and culture consultants. [00:15:25] it's probably a dirty secret, the analytics world, generally the most, dashboards don't really get used that much. [00:20:00] There's this virtuous triangle... You can't have good reliability and good productivity without good safety. [00:28:40] advice I give to everyone that wants to be a data scientist is learn SQL before you do anything else, learn SQL. [00:33:22] This idea of the citizen data scientists become much more prevalent.

Implementing an end-to-end AI product requires a variety of skill sets, especially if the product has to take on a physical form.  IOT engineering, building the AI models, data engineering, and compute optimization are some of the technical challenges involved. At the same time, designing for success needs to take on the tougher challenges of working with the eco system and at times, even shaping it. An AI startup, Vulcan.ai has dealt with every one of these issues and can share with us the reasons why such ambitious projects might fail.     Manik Bhandari and Kamal Mannar, Singapore    Manik Bhandari was the ASEAN Leader for Analytics at EY, based in Singapore. He has been a Management Consultant with Accenture and EY and was part of the founding leadership teams. He has worked with the Singapore Government, especially through EDB to set up multiple Digital and Analytics Hubs and Innovation Centers which has contributed more than $100 million to the GDP. He also serves as the Vice President of the Business Analytics Chapter of the Singapore Computer Society and actively contributes through his business network to build local capabilities in the Singapore eco-system. Kamal Mannar is an accomplished AI leader with experience in innovating and bringing data driven solutions to industries in the public and private sectors. As head of AI in Vulcan-AI he leads a team in developing innovative products that leverage the latest in AI, IoT and data sources like remote sensing to create breakthrough capabilities. He has led development of innovative capabilities like deep learning, reinforcement learning, computer vision and optimization.  He has multiple patents and publications related to application of AI in healthcare, prognostics, health management of assets, smart grids, and precision agriculture.   Highlights:   [00:04:52] There's a Venn diagram that said an ideal AI person has software engineering skills, AI skills, business acumen. It's not possible to have all the skills in one person.   [00:15:07] You can take a few moonshots when you know that your cashflow is okay.    [00:16:40] We want to embed our AI algorithms on a very small footprint.   [00:27:40] You have to build your model from scratch because the requirement is not just accuracy. Is how much power is it consuming? And how long does it take to run on an edge device?   [00:28:35] The idea is you're not building the hardware for the AI, you are custom purposing the AI to the hardware.   [00:29:11] How many flops, which is floating point operations per second, is each model consuming, how much power is it consuming? And therefore, how do I optimize it?   [00:33:54] Basically we were tracking, this is our target cents per hectare.   [00:37:04] Every piece of code is always profiled, not just in terms of the accuracy, but in terms of run time, in terms of logs, for consumption and so on.   [00:41:10] ML engineers: these they're more, pure software engineers and especially because the data we handle is quite huge, it's streaming data, it's images.   [01:10:33] Companies will start differentiating their AI solutions based on, a green mark platinum or something based on the energy consumption.   Resources:   Vulcan AI’s website: https://www.vulcan-ai.com/

The Internet of Things (IOT) has been poised to deliver an unparalleled surge of productivity growth… perhaps for too long? Analytics is foundational to IOT, and the adoption of IOT holds the promise of unleashing a new age of pervasive data. How do we make sense of the hype around IOT and what it can actually achieve? As a highly respected analyst, Phil is uniquely qualified to talk about the potential of IOT and its avatars in the Factory of the Future and in Smart Cities. With Phil’s deep experience in hands-on modeling, this episode provides a two-track insight as he talks about the impact of analytics, but also his own analytical approaches.        Phil Marshall, New Zealand   Dr. Phil Marshall is a co-founder and the Chief Research Officer of Tolaga Research, where he supports thought leadership and product development. Before co-founding Tolaga in 2009, Dr. Marshall was a Vice President at Yankee Group for nine years, leading its global service provider technology practice. In addition, Marshall has held various management and engineering roles at Verizon International, Spark NZ, and BHP NZ Steel. He has over 25 years of experience in the ICT industry and has supported projects in over 40 countries. He has a Ph.D. degree in Electrical and Electronic Engineering and is a Senior Member of the IEEE.     Highlights   [00:05:00] “I think the art in analytics is underestimated. In effect for that matter, I would say the art in mathematics is underestimated… think … how art would be treated in schools, if you treated it more like mathematics and it's almost like an artist would end up spending all day painting fences.”    [00:12:00] “We've had challenges, right? But it doesn't mean that the IOT market is a failure. Absolutely not. I see a huge role for it more broadly, if you'd think about it as the instrumentation of things.”   [00:15:00] “With predictive maintenance 99.9% of your data is irrelevant. It's just that 0.01, which is incredibly valuable when you know, something has gone wrong. And so you have a data sparsity problem, right?”    [00:23:00] “Nine times out of 10, the forecast is wrong. It's not what the number is, it's what the inputs are, what the assumptions are, how you built it…”    [00:26:00] “There's a real balance in automotive manufacturing in terms of where you place that automation, where you place the digital intelligence, and how you develop it.”    [00:31:00] “There are some aspects of manufacturing… where it just makes no sense to automate… So there's a real art of being able to understand where [advanced technologies] are most appropriately implemented and where you draw the line.”   [00:35:00] “To progress smart cities, think of the day in the life of a typical citizen and where are the areas or what are the things that are important? I think the priorities should become pretty clear when you do that.”   [00:41:00] “I underestimated the power of status quo inertia around traditional solutions, risk aversity, stakeholder priorities... it was pretty much across the board. I'd missed all of those.”      Resources Tolaga’s website: https://www.tolaga.com/

Managing today's gargantuan volumes of data has its challenges. These include silos, regulations, data, quality, Extract, Transform, Load (ETL), pipelines, storage, and so on. While organizations recognize the importance of investing in data engineering solutions and embark on ambitious data warehousing or data laking platforms, many organizations flounder in the data deluge. The problem is, engineering doesn't have all the answers. Very few organizations can claim to be in command of their data journey. In this podcast, we try to understand why. Raghuram Bhatt gives us an inside view of the challenges and what it really takes to solve them in the banking and financial services industry.     Raghuram S Bhat, Singapore   Raghuram leads Cognizant’s Banking & Financial Services Consulting practice for ASEAN countries and is based in Singapore. He has worked extensively with regional and global financial institutions across multiple markets in APAC, Europe & USA. He has advised clients on topics related to large scale business and technology transformation programs, organization design, IT & digital strategy, business process reengineering and data & analytics. He has more than 18 years of experience and a proven record in C-level client management, consulting sales & delivery, people management, thought leadership and P&L management.  The opinions expressed within this podcast are solely the author's and do not reflect the opinions and beliefs of Cognizant.   Highlights:   [00:02:53] Traditional banks have made significant investments, but very few have succeeded in diffusing and scaling artificial intelligence and analytics technologies throughout the organization.   [00:05:46] Failures are a gold mine of information, but there are no incentives doing a deep dive on the failures in most banking and financial services institutions.   [00:08:33] Several of the clients that I've worked, success has taken longer than anticipated. It happens when data and analytics is not native to your Genesis.   [00:10:38] I follow radical gradualism, which is have a vision, but take small steps towards that direction.   [00:12:38] Failure's not being dramatic. Several platforms that firms have invested tens of millions of dollars, nobody frankly uses.   [00:18:33] The data and analytics strategy is often not owned by one person.   [00:22:45] Data from the new platform does not reconcile with the old one. So using the new numbers from this platform means you having to restate some of your earlier performance numbers and financial reports.   [00:34:40] Be it the art or framing the problem, there's generally a lack of understanding about what and how data can help in running businesses.   [00:37:04] The ambit of engineering is much broader. This is a fundamental capability required to industrialize analytics.   [00:39:24] Have a data lab, but also take the output from the lab and industrialize it.   [00:40:27] The future of finance is underpinned by deep technology and data and analytics is at the core of it.

Stora Enso, headquartered in Finland, is the renewable materials company with a revenue of about US$ 10 billion. The oldest preserved share certificate in the world dates back to the year 1288 and is of Stora Enso. This makes it, probably, the oldest limited liability company in the world. What can a company like this have to teach us about analytics? Plenty, if you listen to Marko Yli-Pietilä. In this episode, Marko talks about using analytics for smarter manufacturing. Where do product ideas come from? How do you develop a pipeline of successful use cases? And what do failures mean? Listen for very generalisable insights from a very unique setting.        Marko Yli-Pietilä, Finland    Marko Yli-Pietilä is an experienced Digital Transformation professional. As Head of Smart Operations at Stora Enso, he works on digitalising operations to improve operational efficiency. Prior to joining Stora Enso, Marko has held international positions at Nokia and Teradata. He has also worked with renewable materials companies and their digitalisation related development programs for years as a management consultant. Marko holds a Licentiate of Technology postgraduate degree and he is also a certified vocational teacher.     Highlights:   [00:02:00] We utilize modern technologies in a sensible way to improve our operations efficiency. OEE, overall equipment efficiency is the most important KPI set we use to measure success in industrial digitalization.   [00:10:00] We have nearly 40 solutions in smart operations that are operational at least in one mill. And  roughly 20 of those…we are scaling out to multiple mills even globally.    [00:15:00] We can predict 10 to 12 hours into the future how the quality variations will be.   [00:26:00] There are so many ways you can utilize modern methodologies to find a solution. There's not one or 10 or even 100, there might be like thousands of different alternative ways to come to the right solution.   [00:36:00] No matter how good a solution you have in your hands, if the end-users are not happy to use those tools, the value will be zero or even negative.   [00:38:00] One approach doesn't fit all. You need to be ready to listen, to adapt, customize at all levels: solutions, your methods… all that you need in order to take industrial digital forward.   [00:40:00] If you're not having these sorts of cases [failures], then you're doing something wrong. You're not kind of challenging yourself enough. You're not doing, let's say, crazy enough things if you always succeed.    [00:40:00] It is iterative and sometimes you take one step back, but after that you might take three steps forward.     Resources:   https://www.storaenso.com/en Stora Enso website   https://seeds.storaenso.com/ The Stora Enso Expanded Design System   https://hyperight.com/dair-awards-2021/ The DAIR awards

Better decisioning in the “Quote-to-cash” cycle means a lot to the CFOs of the world. Analytics-led efficiency improvements in this cycle have the potential to make bottom lines better. While modelling is a part of the analytics, translating model insights into action often remains a challenge. Debi Guha is a seasoned finance professional and she has packaged her extensive experience in managing the cash cycle into a service on the cloud. She talks about using analytics to get better at the abstruse problem of collecting and managing money.     Debi Guha, Singapore     Debi Guha is an Enterprise SAAS company founder building tailored analytic solutions for CFOs to navigate an ever-changing world. She transitioned from the private equity and private debt investing worlds to set up TwoDotSeven. Today, her company is providing AI/ML led solutions to help solve wicked problems in working capital including credit risk and demand forecasting for inventory management.      Highlights: [00:04:00] If a company can successfully reduces net working capital, by about 5%, every year in a five year timeframe, without any external capital, you can potentially release enough cash to grow your sales four times and your valuation eight times.   [00:05:34] we are an analytics first software company. Predictive analytics creates focus, which are then fed into appropriate decisioning grids that are then actioned through our solution.   [00:22:00] We typically create multiple grids, for different customer segments for different geographies, because what the grid has to reflect is, what is the company's strategic intent?   [00:24:36] During COVID last year, most companies decided to extend, their DSO appetites, because they realize that companies paying abilities have got handicapped.   [00:30:31] the terms of engagement- the contract, what are the terms that they're signing on has a significant, bearing on what their payment behavior is going to be.   [00:39:56] today you have significantly better data and tools at your disposalto do a superior job.   [00:40:42]  providing forecast is not good enough, but creating a framework to which this can be implemented becomes important.   [00:44:15]  What, however changed, was the line. If the best customer was a seven day payer, he slipped to becoming a 15 day payer.   [00:44:31] last year during the pandemic, when we were creating grids, the cutoffs had to be moved higher.   [00:45:04] And short, predictable cash conversion cycles, have become paramount because of COVID because everybody was running short of cash.     Resources:   Two dot seven website: https://twodotseven.com/

About 10 years ago, a local authority in the UK developed a really useful database of the local population to identify people who were at risk of being admitted to hospital. When they got down to using it, though, people didn’t seem to be in as acute a level of need as they expected them to be. What went wrong? Douglas helped them understand: “I don't know how long it would've taken them to figure out that that was what was happening. But the model brought them to that conclusion much faster.” Health policy is fashionable in the post-pandemic world. In this episode, we feature insights about analytics from someone who has been modeling health systems since 2002.       Douglas McKelvie, United Kingdom   Douglas McKelvie has had a varied career, starting as a social worker in 1980. In the 1990s, he worked for the UK regulator of social work education, where he took on responsibility for workforce development, whilst his further studies introduced him to modelling and simulation. In 2002, he moved into consultancy where he worked alongside Eric Wolstenholme, from whom he learned in depth about the System Dynamics approach. Douglas uses simulation to support strategic planning in health and social services. He has built simulations for a wide range of organisations across the UK, including government departments, NHS bodies, and local authorities. He mostly uses System Dynamics, in a group model building context, an approach that takes a ‘whole system’ view of complex problems, such as are found in health and social care integration, where services span agency and functional boundaries, and in children’s services. Models include delayed discharges, commissioning and capacity planning, workforce planning, economic impact analysis, public health / population level issues, child-care workforce, and children’s mental health. Douglas is the Past President (2016/17) of the UK Chapter Policy Council of the System Dynamics Society and has written about the use of simulation in health, social care, and related human services. He is Edinburgh-based but works across the UK (and beyond). See Wolstenholme E and McKelvie D (2019) The Dynamics of Care – Understanding People Flows in Health and Social Care (Springer)     Highlights:   [00:04:00] The kind of people who work in health and social care are generally very serious about what they do. They do it and they do important work and they, and they want to do it well.    [00:14:00] When you're dealing only with data and not with structure, you don't necessarily know what questions to ask of your data.   [00:20:00] We see something that's needed. We meet a need. That's a good thing. And in meeting that need, maybe we increase the level of need, and that's also a good thing. But it does make us pause and think, well, what kind of systems do we need, uh, in order to look after people properly?   [00:26:00] It seems to me implausible that in the realms of complex social systems that there is some kind of killer equation out there that's going to give you the answers.   [00:29:00] So sometimes I look at papers…where I think that somebody spent a lot of time trying to understand a problem and then has reconfigured that problem as some incredible equation that means something only to point not not not 5% of people who might read the article.   [00:33:00] I think there is an art in involved in trying to figure out what is going on in the client's mind and how might a model help.   Resources:   Symmetric Scenarios website: https://symmetriclab.com/    Angele Peters’ Ph.D. thesis: https://pure.uvt.nl/ws/portalfiles/portal/1540305/Thesis_Angele_Pieters_August.pdf   Geoff McDonnell: https://preventioncentre.org.au/people/geoff-mcdonnell/    Whole Systems Partnership: https://www.thewholesystem.co.uk/employees/peter-lacey/    The Health Foundation: http://health.org.uk/

A lot of analytics is about who, what and when, but the question of why has not received a lot of attention. “Correlation is not causation” is an old aphorism. But then what is causation? Ajith Sahasranamam created his company, Ongil (the name means Dolphin in Tamil), to focus on causal analytics in the consumer space. In this episode, he talks about his work in applying analytics to suggest specific actions for clients based on causal understanding.      Ajith Sahasranamam, Salem, India   Ajith is the founder and CEO of Ongil. He helps data driven decision making possible for enterprises during uncertainty. Ongil serves Fortune-500 enterprises to make effective decisions for dynamic e-commerce scenarios. Ajith is putting together a team of data scientists and data engineers who challenge conventional approaches to data science. He has a PhD in Computational Neuroscience and is an avid reader.     Highlights:   [00:05:28] we don't rely on the intelligence data of a company. we try to bring in external data that captures the dynamism of the market. But the insights can only go so far.   [00:06:01] Especially in a multi-dimensional space, what one needs is specific, action points. Not necessarily insights.   [00:16:56] I remember reading a book. It was called big data at that point in time, it basically had a chapter saying that, causality is dead, correlation is a new science because of machine learning. But now we know that it's not true.   [00:16:05] The common example is the chocolate consumption in a country is directly correlated to the number of Nobel prizes. This is not as spurious as it sounds.   [00:22:04] for something to be called causal, two events cannot happen at the same time, because they still need time for this event to influence another event.   [00:30:44] What causal modeling does is it makes it explicit instead of living with implicit assumptions of how things are related to one another.   [00:40:40] In any kind of modeling, domain expertise becomes important. And more so in terms of causal modeling.   [00:46:40] When doing causal modeling , I spent most of my time framing the right question. Once we frame the right Question in getting to the answer, I wouldn't say it's very easy but easier   [00:56:14] We want to become the search engine of e-commerce insights.   [01:01:27] People make up these stories , WhatsApp universities, at least saying that Muslims men, they typically tend to marry four women always and so on…     Resources:   Ongil website: https://www.ongil.ai/

Welcome to The Art of Analytics podcast. We talk about the things that don’t get much air time. Isn’t it more important to choose the right questions, and to frame them right, than to know how to compute the answers? To know where to go to, than to get directions from maps? Because we worship success, do studiously avoid learning from failures?