9.3% of the checkout records have been excluded from the results! It is because the associated information about these records is not available.
A barcode is attached with an item, and each item in the checkouts data is associated with a barcode. Barcodes come and go, and the item itself may also not be available by the end of the year. The list of items is only the list as of the end of the year, not during the entire year (as one would typically expect).
In 2013, out of 8,289,052 checkouts, 772,975 of them do not have associated catologue information as available from the open data source that we use. In other words, the checkouts analysis shown in this site is skewed towards the 90.7% of the data available.
Much of the data presented here is mostly based on branch "checkouts", which is different than branch "holdings".
For example, if an item is checked out at the Carlingwood branch, but the items itself is from the Main branch, it is tabulated as a checkout from Carlingwood, not Main. In addition, all other related information about the book (title, media, subjects, etc) will be correlated to the Carlingwood branch, unless it is specifically indicated otherwise (e.g., "home branch").
The gaps in the checkouts densities chart are due to weekly branch closures, holiday closures and other closures.
The source data is in MARC 21 format, which required deep understanding of the library catalogue system. The parsed results will have errors, especially when associating items to their appropriate media types. This may improve over time.
The checkouts data also contains records from non-branch outlets, such as Hunt-Club CC, Collection Management, etc. These checkouts represent less than 0.8% (62,357/7,516,077) of the checkouts with known associated catalogue info. For the most part, they have been excluded from being presented in the app.