Speedier Message Viewer Analysis with IRIS Whiz

As an interface developer I often get asked questions that require investigations into large quantities of messages. For example during a recent meeting our project manager asked me how many sites were actually using our newly set up orders interface.

Usually I'd be trying to copy the Message Viewer output to paste into Excel or simply run a message report for each individual site placing orders and using the message count returned…

This time however, using the Iris Whiz browser extension I had options.

 

Option 1 - Simple: Export CSV

An idea implemented from the InterSystems Ideas portal, simply click the Export as CSV button in the IRIS Whiz button bar to download the current search as a CSV file for easy Excel/Sheets manipulation.

 

Option 2 - Fancy: Analyse

In this case, I had just completed the Analysis tool in my Iris Whiz browser extension.

By adding the PV1-3.2 value to my message search criteria in Message Viewer I could easily run the report, click analyse and instantly have this information to hand in a simple doughnut chart - no exports needed.

 

 

Next, the project manager wanted to know what types of exam these sites were ordering. I added the OBR-4.2 value to my search criteria and re-ran the report. Clicking the analysis button now showed me the sites ordering and the exams ordered. (Each message search criteria is presented as a doughnut graph, tagged on to the end of the graph half of the analysis page)

Queue the third question.

Which sites are ordering which orders?

By clicking the required site in the interactive Doughnut graph I could view the data in the Data Viewer half of the analysis page. Another click on the filter button inside this box applies this data selection as a filter to all graphs - meaning that the exams Doughnut graph now shows the exams ordered for this site only.

Site graph and exam graph filtered by site:

 

And finally the hardest question.

When is this all happening?

Sifting through message times in the Message Viewer page to see when orders are being placed is a non-starter...

Fortunately I'd added a timeline graph to the analysis page.

I removed the filter and clicked the 'Show on Line Graph' button (toggled to ‘On’ for the PV1-3 chart in the screenshot above) to show the site data on the timeline graph at the top of the page.

A quick screenshot later and we were able to send out this report to our sites so they could confirm the number of orders for each day and ensure everything was working as expected.

These reports were to be run weekly, but luckily for me this task had become easy, especially when paired with the saved search function in the message viewer page so I never had to remember which search criteria to add.
 

Final Points

1. Sensitive Data:

The data in your Message Viewer search is sent to a new browser tab and as soon as the tab is closed it is gone - so no worries about sensitive data being saved in the browser. If you want to save a report use the default InterSystems functionality for Saved Searches and just run the report again at a later date. I had planned on a saving mechanism to save searches from the Analysis page but it didn't make the cut in this version.

2. Speed:

The analyse page is powered by the message search and I’ve put no hard limits into the amount of data it can show. The more messages and search criteria you add to your search, the slower the analyse page will go. With that in mind I’ve added a pop-up if you try to load more than 200 messages which allows you to choose if you want to load the bar chart at the top of the page or not. 

The bar chart shows every message as a selectable box. Clicking the box on the chart will add the message to the Selected Messages box in the data viewer (left) side of the page. You can then click the ‘View Selected Messages’ button to open these messages on a new page and take advantage of the message comparison features of the extension.

When clicking this button try not to have too many messages selected. Up to around 10 should be fine. 

Loading the bar chart with large data sets of 10,000s will definitely not be good for your browser but I've left it up to the user to decide.