We were recently invited to IBM to get our hands dirty under the bonnet of Watson Analytics. So analytics consultant Matt Jamison rolled up his sleeves to discover even more about its power and wider potential. Matt (being Matt) focussed on how Watson Analytics could be used to help our clients… We, in turn, wanted to share his thoughts with you. 

I recently attended a workshop at IBM on Watson Analytics. Not only was it a chance to really get to grips with the tool, but *listen up BI guys* it also gives a flavour of what to expect from the new Cognos business intelligence release – the “Assemble” part of Watson Analytics has a common look and feel to the UI in Cognos Analytics (for that reason alone it’s worth a proper look).

But what about Watson Analytics as a tool itself? And how might it help customers?

Despite some reporting and predictive analytic capabilities, its not envisaged that Watson Analytics will supplant anything in BI or SPSS. It’s pitched as something to complement existing tools to help end users really get an understanding of their data.

Attractive visualisation

In its simplest form, Watson Analytics is a starting point for customers wanting to explore and analyse their data. It’s heavy on attractive visualisation and aimed directly at business users. Where its strength lies is in its ability to apply some basic statistical/ predictive analysis to a dataset very quickly and very easily, with more advanced and powerful predictive algorithms to come. I could definitely see it being used as a self-service tool for customers to use over the top of pre-fab queries.

There is limited integration with existing tools at the minute, although that should change. (It already enjoys integration with Twitter, and the recent tie-up with The Weather Company was quickly put to good effect, with the inclusion of meteorlogical data demoed in Watson Analytics at the Insight 2015 event in Vegas.)

Currently you can import data direct from a Cognos report – but it has to be a basic list report – and you can connect to corporate data sources (such as DB2 or SQL Server) but you can’t use an FM model, or anything like that.

You can’t export content from Watson Analytics back into Cognos at present, but you can export the predictive models it creates and import them into SPSS for further analysis.

Dashboarding

Its looks very easy for a customer, given some data, to knock up a dashboard with objects that are actually populated with data. This would allow them to interact with their data and immediately get an idea of the value of what they are asking for. This could then be used as a spec to go off and create something that’s more tailored in terms of interactivity, functionality and branding that can also be put into regular production.

I could see this cutting down on issues like re-working elements of a report because a development hasn’t quite matched the customer’s expectations, for example, or just generally upping engagement.

Mobile dashboarding on Watson Analytics

Washboarding on Mobile devices Watons Analytics

Lastly, it looks like IBM are really championing the tool.  They have a busy development plan for Watson Analytics, it’s fully supported, has community support and IBM has a team of “evangelists” dedicated to helping partners and customers get the best out of it.

The Future?

It’s clear that IBM are billing Watson Analytics as a game-changer to the data discovery and insight industry… It has the ability to automatically present a number of key insights or questions that could be asked of any data set, while allowing business users to ask their own natural-language questions of the data too!

Looking to the future, the ability to get data direct from TM1 cubes and more Cognos sources (like FM models) – and then being able to export content back in for further, more detailed development – would essentially make this another “studio” tool that would be quite powerful. As it is, it’s a tool with some obvious potential uses but also some limitations too.

I’d urge you to have a play, though, and see what you think.  If you want to ask me anything, feel free to get in touch.

Posted in BI, IBM, Watson Analytics On November 12, 2015 By