Acting on the Findings from Discovery Analyses

This is the continuation of the transcript of a Webinar hosted by InetSoft on the topic of "The Topology of the Visualization Vendor Landscape." The speaker is Abhishek Gupta, product manager at InetSoft.

How do users act on the findings from discovery analyses? How do users act on relationship findings from discoveries that may require further data analysis software implementation work or require further visualization? So in our world, you do the discovery and analysis, and you can do several things when you reach a selection state you’re interested in. You can export the charts into PowerPoint, Word, PDF, or Excel, as we saw in the one example.

You can export also the rows and columns from any of the tables and memories out to a .csv file which can then be imported into anything else. So we do have customers who will link the analysis results to their other reporting systems.

We have customers that have linked these things through to control systems so, for example, if you find, say problems and you want to pass that information back to your planners, you can connect that right back into the operating systems to do that. So I think that’s answer to the question. We’re trying to be a discovering analysis tool that connects with the rest of the world you operate in whether it’s Microsoft Office so you can make presentations or whether it sending data to another system.

A lot of times the output will be a list of customers that goes back to a mailing system. And maybe, we just export the ID numbers so they’re keyed in, or maybe we export the emails. What’s in the export is just determined by what’s in the base underlying data tables and what the user actually wants to export. There’s a reason why you might not want to export e-mails.

Maybe you don’t want the user to direct e-mail. You want them to go through your mail engine and use all the rules in it. In other cases you might want to export the e-mail so how we interact coming out of an analysis, it can be many ways.

Here’s another question on the predictive modeling, how hard is it to use? What we have done is taken really rich regression algorithms from a leading player, and we put them in a very simple interface, so that you do not need to know modeling to use them. You just need to have an understanding for the data.

And then that regression I literally ran probably 25 models behind the scenes to pick the one it picked. It selected the one with the best fit, and then displayed the output. The other thing with all the data in memory, it’s easy to make the selection visually. The modeling picks that up. If you have done scoring, you can look at the scored population, and it will go back to all the other charts.

So you can quickly explore the scored population in the map. Where are the high scorers? What are their ratings? Are they staffed, or they’re unstaffed? That’s really cool to be able to take the model output and without going to another system examine it visually across all the normal fields. The third point on our predictive modeling is often, you have got a table of people, and you have got the address and some demographics, but what’s not in that table is their prior purchase history or the fact that they’ve, responded to the three promotions that are in the promotions table, or they’ve come to six events that’s in the events table.

Well, in Style Intelligence it’s really easy to mash up these other tables. Then they’re available for the modeling, and that’s a five minute exercise without touching the database which then gives you these factors from all these tables to run in the model. This is an advantage of these kinds of lighter interactive discovery analysis tools. If you had to do that and go back to the IT group to get the database tables change, it’s a much bigger deal.

View a 2-minute demonstration of InetSoft's easy, agile, and robust BI software.

Some people call this an analytic sandbox where you can play with these things very easily on your own. Now if you actually are going to do this all the time, sure it should be in the database, but often at the beginning you don’t know, and this is a format you can play with the data in a self-service mode.