This is the latest in a series of guest posts by Greg Barniskis on the status of ongoing ILS reports enhancements.
Code Clustering Subtotals
The work on code cluster subtotals, first published last month, continues to be experimental. Please do not rely on these subtotal numbers until they are announced as being stable and accurate. The county cluster subtotals on June's reports were definitely not accurate (despite our successful computer simulation of it, Fitchburg is not actually located in Sauk County).
For now, the cluster subtotals feature is still evolving and is presented only for the sake of discussion. This month's reports feature new PSTAT clusters, of the kind we'll need to have when we do the state's annual reports in February. Sorry, there is still not much in the way of documentation for the clusters, they need to stabilize before that can be done in a meaningful way.
New Holds Reports
Other recent reports work has focused on a new format for the main holds statistics report, and the creation of several new holds-based reports, including:
- Holds placed by time of day
- Holds placed by PSTAT
- Holds placed by PTYPE
- Holds filled latency (days to fill)
- Summary of most-requested titles
You will notice that the main holds statistics report now has many more columns than last month, perhaps even too many. Like all new and revised reports, these new columns are presented for your review, and the format and content may continue to evolve based on your feedback. Many new columns are self-explanatory, but some may not be. Column notes are also being developed.
You may also find that the holds statistics within some columns look rather different than comparable columns from previous months, having numbers that are notably larger or smaller in some cases. Work on detailing all of the new holds reports has revealed that certain assumptions being used in the past were not entirely correct. The new breakout format makes some different assumptions and thus arrives at some different answers.
In particular, the "holds placed" counts for all previous months were unintentionally dropping counts for any holds that were both placed and cancelled in the month, or any holds that were placed, trapped, and then expired on the holds shelf. The new report format properly accounts for these canceled and expired holds, resulting in larger counts for total holds placed. That's the bad news (that older counts are off), and the good news (future counts will be comprehensive).
The best news is that a breakout of holds filled latency (days to fill) tells the story of how 23% of all holds filled in July were filled within three days of being placed, 58% were filled within seven days, and 75% within three weeks. I am duly impressed, and my hat's off to all of the library staff who make that happen. Nice job!
Beyond three weeks, the holds filled latency report shows that the hold fulfillment rate tapers off quite slowly. In part that's due to the influence of popular titles with long queues, and in part it's due to patrons choosing to suspend their holds indefinitely. The oldest hold filled in July had a latency of just over four years, representing a hold originally placed and suspended on Dynix in the summer of 2008.