- Please tell us about your educational background
- D., Chemical-Environmental Engineering, UT, Knoxville, TN
- F.A./MBA, Institute of Chartered Financial Analysts, India. Silver Medallist.
- MS, Chemical-Environmental Engineering, Johns Hopkins & Clarkson Universities,
- BS, Chemical Engineering, Indian Institute of Technology, Madras, India
- Please tell us about your professional background
- Proven CXO & Founder/Leader
- Founder & Chairman, Sriya Group, Renmatix, Inc
- President & CEO of Reliance Bio-Refinery
- 2 successful US IPOs with over $100MM raised including “Green Shoe”
- 2 Successful M&A with over 9X return to initial investors
- Chief Knowledge Officer of Tata Chemicals (>$25 billion global corporations)
- Expert in prescriptive analytics, predictive analytics, machine learning, math-stat algorithms
- Expert in Supply Chain, ERP, Manufacturing, Financials, BAM, DSS
- CXO of Commodore Separations, Inc. (CXOT)
- Please tell us about your company and big data products
- OUR VISION
To be the GLOBAL benchmark tool for quantifying and enhancing customer Digital Experience.
- OUR MISSION
To enhance customer digital experience (DX) resulting in improved market share and profitability for businesses that implement predictions and recommendations based on Sriya’s custom indices – DXi & DXi+
DXi quantifies and summarizes big data and analytics into a simple overall index and a set of sub-indices for selected categories.
DXi can be measured for the organization as a whole, for each business segment, product lines, and/or selected profit centers.
DXi predicts business outcomes that are tied to ROI, sales or other user’s specified criteria.
DXi+ is the Machine Learned Experience index that uses advanced machine learning algorithms to identify key DX variables.
DXi+ prescribes multiple decision paths (Decision Trees) for enhancing Customer Digital Experience.
DXi+ is a self-learning model that measures DX, predicts business outcomes and prescribes decision trees with increasing levels of accuracy with each successive iteration.
- Who are your major competitors?
There are no direct competitors to our DXi services (providing a unique Digital Experience Index) but there are other companies like kissmetrics, Foxmetrics and Predixion which provide only part of the offerings in capturing the interactions of users on the website and could be our potential partners.
Our tool would complement Google and Adobe analytics tools by adding more intelligence to their data through the DXI score and its predictions and prescriptions.
- Competitive Advantages
- Simplification and summarization of voluminous data/analytics into an easy to understand index (DXi)
- Accurate and actionable metrics, predictions and recommendations based on statistical and ML algorithms
- Ability to track individual as well as aggregate customer experience information to gain deep insights into digital behavior.
- Self-Learning, clustering and classification techniques enhance the model’s predictive capabilities and make recommendations more reliable over time.
- Your views on why digital experience is so important
- Current digital interfaces and mobile apps have a more than 65% failure rate due to poor customer experience.
- Present day analytics lead to more complexity than clean solutions; they don’t always consider impacts on ROI and profitability.
- Current solutions are based on low-response surveys and are not representative of DX.
- Despite this obvious problem, there are
no comprehensive methods to capture the
customer digital experience and translate into measurable indices and recommendations that can be used to strategically improve the bottom line.
- There are so many companies performing digital analytics like Adobe. How do you compete with such big companies?
We do not compete against existing analytics products. Rather, we complement and enhance products like Google Analytics and Adobe. While they focus on data visualization, we focus on converting our data into DXi and DXi+ using our proprietary architecture.
We also support API to plug into Third Party analytics data like Google or Adobe Analytics so that we could process their data and convert into our unique DXI score with predictions and prescriptions.
- Do you compete with big companies like IBM or SAS on big data platforms?
We do not compete with IBM or SAS on Big Data Platforms but our DXi tool will be a complement to these existing Products and will be a key enabler in the Big Data space by helping to identify critical few from Trivial Many.
We could join hands and collaborate with big companies to add more intelligence to the Big Data.
- What are your views about future of big data?
Big Data is definitely going to get bigger not just by volume but also by variety. That demands more intelligent ways to slice and dice data and to make it easier for the users not just to visualize but also to interpret to make meaningful decisions. We definitely need to present some pointer solutions for the users rather than putting the entire onus on Data Scientists to come up with predictions and prescriptions.
- Which verticals do you support right now? Healthcare? Finance? Media? Retail? Oil and Gas? etc..
Retail, E-Commerce for now is our focus. We have domain expertise in Chemical, Oil & Gas
- Please tell us in detail about your machine learning algorithms and what makes them better as compared to other algorithms in the market.
It is our architecture consisting an ensemble of many individual ML algorithms that we use, make us unique.
Our uniqueness, and why we have value, is due to:
- the type of data we collect which looks at every individual visit as an unique record and not just data aggregation,
- Integration of the data into a proprietary math-stat model which converts this “Big Individual Data” into a quantitative index/score (DXi) a number that we believe not only represents but quantifies an individual’s subjective experience on a digital portal.
- Finally, the use of an ensemble ML algs plus some basic statistics and a simple voting algorithm to combine the output from the ML algs with DXi, to generate personalized predictions and prescriptions like decision trees, sales predictions, ad campaign lists, action items etc
Right now, we are working only on websites, but extending our technology to mobile apps is the obvious next step.
The data collection occurs during the actual visit to the website by an individual; the data does not come from a post facto survey or other method that has any intrusive or self-selecting aspect, or is convoluted by other experiences, such as delivery time or experience with a product purchased – neither of which are part of the experience of an individual while they are on a website.
The data we collect is also “precise data” that answers the desired question. This is analogous to the type and amount of data that you collect when you meet another person for the first time; you don’t demand a massive NSA dataset! You ask their name, and then questions that are as relevant as possible to the situation and the encountered you are having. And from that relatively small amount of data you make a decision about the experience you have just had, and can go on to collect more data later.
- What is your advice for the recent graduates who want to venture into the field of big data?
“Big Data” is here to stay and will get bigger. You all will need to rise higher than the “Big Data” stack by being innovative and disruptive. Machine Learning Algorithms will keep getting better and better and could be a great future for all of you. Predictive and Prescriptive analytics is the future
- What is your vision of machine learning for future?
ML is under utilized and misunderstood because it is not easily accessible, so my vision of the immediate future is that ML should – and will – become increasingly accessible for people to use easily, and they will use it in ways that we cannot imagine now. Imagine asking a member of the public in 1900 what they thought the future of the internal combustion engine would be? Or just remember what everyone thought the future personal computers would be when the Commodore 400 and Apple II came out!!
- Which algorithms in machine learning are best suited for digital experience and how your customers will be benefitted?
A great question, and we are still learning the answer ourselves. The list will certainly change as we develop, and as new algs become available, or are shown to have unexpected properties. And of course, we cannot do everything. So we have restricted ourselves the following list;
Random Forests, Principal Component Analysis, K-Means and Support Vector Machines are the best for generating our DXi score, while kNN is clearly suited to sales predictions. We use a Decision Tree algorithm to chart a user’s path through the various data (events) that we collect. We also combine our DXi score with the RFM inputs (using methodology employing Customer quintiles) and use for generating a customer profile for the client.