Embedding Predictive and Prescriptive Analytics
The key to being able to embed predictive analytics is to be able to understand the thought process that a human puts together in coming up with insights and recommendations and structuring it in a workflow that can be programmed. Once that is done, the rest is something that any programmer worth his salt can convert into a code/algorithm that generates the recommendation based on the data.
One technique that is typically used is to create weights for each insight based on how far off the data anomaly/trend is the norm and also allowing for user feedback to over/underweight that recommendation the next time the same thing happens. This weighting by the end-user ensures that the learning from each recommendation is applied to making the bot/algorithm more intelligent.
Another aspect of this trend is the widespread prevalence of bots and most of the largest tech companies, be it Microsoft (with its LUIS enabled Bots), IBM (With Watson), Google and Facebook have started investing heavily in Bot Technology. In fact, Microsoft has come up with a Bot Framework that allows you to build once and deploy many times over. The same bot and logic can be published to Whatsapp, Slack, Skype, Skype For Business, Line and also your Website with just a few button clicks.
Adding predictive analytics to existing reports and dashboards is already widely prevalent. Most BI Tools have off-the-shelf features that help include Predictive analytics into the reports and dashboards. Enabling BI infrastructure and dashboards is an interesting idea. One could think of using historical data to predict what times should certain infrastructure be scaled up / down resulting in Dev- Ops cost savings.
Convergytics has already done work in this space for Telecom companies looking at managing their Network (Tower) infrastructure based on predicted call volume in each cell. The same concept could be applied in distributed hardware setups as well to ensure peak performance and low costs.
Intelligence could also be enabled in dashboards to show the right dashboards to the right customers based on segmenting users and looking at historical behaviour in terms of what kind of dashboards the users may have interacted with in the past. This should result in increased engagement of users and also bring in more intrigue to users and an element of surprise that would make users want to come back and check what their intelligent friend (bot) has to share today.
Is India Skill-ready for Adopting Core Prescriptive & Predictive Analytics
Encouraging more industry interaction at schools is key. One can learn all there is to learn from an advanced statistics and machine-learning perspective. But, applying it to industry examples and getting value from the data is a whole different skill all-together. Another aspect of this that can bridge this gap is to have industry-academia interaction. Convergytics has engaged more closely with a few business schools including the IIMs and this has not only been enriching for the students, but has also helped us get new ideas from the interaction with the students. This must be encouraged and the government could incentivize businesses to do so and also incentivize start-ups/companies that offer internships to young students.
Self-service BI is the way forward. Insight at the speed of running businesses means that having access to answers to questions as they come up is the only way to stay competitive. Gone are the days when a business user created a requirements spec and waited for weeks together to get a report. By the time the report was delivered, the requirement itself had changed since there was a change in the data or even a change in the business dynamics itself.
The role of IT has changed from creating reports to creating a data-structure/data-mart that allows the end-user enough flexibility to get answers to the questions. Another aspect that the IT team can differentiate themselves is enriching this data set with more and more sources so that the insights can be driven by a 3600 view of the customer.