(Article originally published by Zinnov)
In this digital age, with its growing number of customer touch points and proliferation of connected devices, enterprises have access to vast amounts of non-traditional, unstructured data sources such as blogs, social media, videos, and speech as well as real-time sensor data. As the volume and variety of data rapidly increases, enterprises find it harder to derive meaningful insights from their data.
We spoke with Manoj Karanth, Head of Big Data, Analytics, Cloud and DevOps at Mindtree, to gather his perspective on this subject and also learn how Mindtree’s cognitive big data analytics platform, Decision Moments, helps enterprises tackle challenges with big data.
How have you seen big data analytics evolve? Can you describe some key advancements?
The major evolution in the analytics space is the drastic increase in processing speed, which has vastly reduced data-crunching time. Moreover, the growing accessibility of cloud computing has made it easier to derive deeper contextual insights through integration of third-party data sets. For example, we helped a customer optimize stock order quantities by providing insights on how demand is affected by nearby places of interest, population and demographics. We did this by bringing in Google point of interest (POI) data and other demographic data sets. The third important piece is the ability to deliver personalization at scale.
For business success, it’s imperative to derive meaningful and accurate insights from big data. What challenges do enterprises typically face when trying to do this?
The first major hurdle is to make sense of the data coming from disparate sources and derive insights from this data at speed. For example, although creating a single view of customers has been talked about for ages, it still remains a challenge due to the increasing variety of data (social media, blogs). Another difficulty is finding apt third-party data sets and integrating them with internal data. For example, enterprises are struggling to find a platform they can use to combine demographic data with customer data and enable in-depth customer profiling. Doing this in-house requires complex data science capabilities, which points to the third and most prevalent problem—a shortage of skilled data science talent.
External data coming from disparate sources requires cleansing and conversion to a unified format. Do you think that also poses a challenge today?
While cleansing and data conversion are troublesome, the real issue has been around metadata and the ability to scale it (again, because of the growing volume and variety of data). In the analytics space, the start-ups with the highest funding are those that operate in data cataloguing and data wrangling. Currently, most companies store data without any schema and give little thought to the use cases that need to be powered.
Many people believe that the cloud has made it easier to run advanced analytics on big data. Do you see a symbiotic relationship between cloud computing and big data analytics?
The cloud has revolutionized the way enterprises run analytics by making storage independent of processing and thus enabling on-demand processing with dynamic data access. This has greatly optimized data processing in business scenarios where continuous processing is not required. Plus, separating storage and processing makes it possible to choose the technology according to the use case. For example, to create a comprehensive customer view from the high-inflow data, Apache Cassandra can be used for storage, and to analyze the relationship between the data points (when a graph-based view is advisable), Apache Spark can be used for processing.
Mindtree recently launched a big data analytics platform called Decision Moments. Can you tell us about its core value proposition and target customer segments?
In the big data solutions landscape, there is no one major technology platform that can cater to all aspects of big data analytics. Existing solutions require a great deal of sandboxing, where data engineers have to mix and match and figure out which technologies can be used together. The core value proposition of Decision Moments is accelerating the speed to insight. It helps enterprises plumb different big data solutions with high accuracy, minimum effort and fast turnaround times. Moreover, it has pre-built, industry-specific machine-learning models and third-party data set integration capabilities. Decision Moments is targeted at two sets of users: big data engineers and business users. The engagements range from exploring novel use cases for data to addressing specific business problems.
How does Decision Moments differ from competing analytics stalwarts such as SAS, SAP and IBM?
We work closely with major tools and platform providers, bringing in our domain expertise, pre-defined machine-learning models and integration with third-party data sets to deliver meaningful insights. Decision Moments actually complements larger products and platforms.
Does Decision Moments focus on certain industry verticals, or is it a service-line platform that can be used across industries?
Decision Moments has a horizontal focus from a big data processing standpoint as well as pre-built, industry-specific machine-learning models. We use it as our core platform for any vertical-specific solutions. For example, for the retail sector, we’ve drawn on our previous work and capabilities in promotions and supply chain analytics to build outcome-based machine-learning models with Decision Moments as the platform. For the travel industry, we built a “connected traveller” solution that enables airlines to provide seamless travel experiences to frequent travellers.
Does the platform handle advanced technology capabilities such as artificial intelligence and machine learning, which are increasingly gaining traction in the market today?
Decision Moments has mature cognitive computing capabilities both from an interaction and intelligence standpoint. It is integrated with visualization and sight-based interaction, bot frameworks for conversations and machine-learning models, in both supervised and unsupervised states. We’ll soon be integrating deep learning capabilities with Decision Moments.
Can you provide us with a couple of actual customer scenarios and use cases of Decision Moments?
For a leading trust bank, we prepared a churn model to identify groups with high churn propensity using the bank’s customer data and designed a targeted marketing campaign. For a coffee machine manufacturer, we built a predictive maintenance model that enabled the company to add a new revenue stream of providing preemptive maintenance services. A leading Middle-Eastern airline entrusted Mindtree to overhaul its revenue management system, which used to take 24 hours to provide insights on daily financial statistics. Decision Moments reduced the time from 24 hours to eight seconds.
Final word! Going forward, what will be the key focus areas for Decision Moments?
We have two immediate priorities: to enhance the existing self-service capabilities and to enable an “as-a-service” model for the platform.
Learn more about Decision Moments from Mindtree