This was a good session presented by Emily Morton-Owens of New York University and Karen Hanson of the NYU Medical Center.
Emily and Karen described their problem of wanting to make the monitoring of information easier. They had a new director who wanted to know how
the library was being used. They had access to a lot of existing
data and had multiple uses for this data: to make decisions, to
show worth, and to make one-off reports.
Their data came from open source,
homegrown, and available sources. In deciding how to create an interface with graphs showing the data they used “above all else, show the data” and the principles of Edward Tufte to guide them. It was important to have
truthful propotions in the data and eliminate graphical junk. For this reason they eliminated the use of pie charts, which have been determined in studies to be sometimes difficult for users to properly interpret, and instead used bar charts.
They built their bar charts using Google charts, which made the process quite simple (they showed the entire code for a chart that they had to write and it was less than 50 lines). They also implemented a nice linear regression line and described how they did that.
The dashboard they created has multiple graphs on it that are not necessarily related except that they show information about the utilization of resources at their campus. They used as a principle that a good dashboard should have benchmarks and goals. The dashboard that they wound up developing and demonstrating was geared toward providing an operational view of their data, although they would like to adapt it to show a strategic view.
One of the things that they discussed in their data usage an analysis was the parsing of EZProxy data, which I use a lot. I haven't seen many presentations on parsing EZProxy data before, even though I'm doing it frequently, and it was interesting to see their approach.
No comments:
Post a Comment