Print

Print


At UW-Madison we have a mixed environment. Our campus has invested in a Tableau Server instance, so we participate in that environment. We have a single viz/dashboard describing our library instruction activities:

https://dataviz.wisc.edu/views/UW-MadisonLibrariesInfoLiteracyInstruction_0/HomePage?:embed=y&:showAppBanner=false&:showShareOptions=true&:display_count=no&:showVizHome=no

For our internal facing usage statistics, we tend to host our own web pages and apps that utilize the D3.js library for data visualization. Here are a few examples:

Collection Statistics
https://web.library.wisc.edu/sp/cca/

* Basic stats mostly about the size of our collection and its use
* Data spans over a decade across to ILSes (Voyager and Alma)
* Some viz's are used to tell our story to campus admins or were even requested by a provost [1]
* Others get into an insider baseball view that is intended for library staff with knowledge of operations [2]

Journal Statistics
https://journalusage.library.wisc.edu/journals/991021974405202122

* A focused look at serials use
* Combines print and electronic usage for macro trends
* Rails backend

If you will permit a bit of a longer reflection on this space, personally, I think there are some aspects of using the newer GUI tools like Tableau (or Microsoft's product, etc) that are not talked about enough. Data visualization is an extremely complex process. It involves:

* formulation of good analytical questions, 
* sophisticated knowledge of data sources and their associated formats, 
* data collection and processing techniques and 
* presentation techniques for quantitative data (including aspects like color theory).

Tableau is good, opinionated software that attempts to encourage good practices over bad ones. And it can generate some very impressive data viz. But from a software development perspective, there are some highly problematic elements akin to the difference between managing data and computation in a tool like Excel vs., say, a Python project in a Git repo (a la the Software/Data Carpentry practices).

For example, for other software engineering processes that are as complex as data visualization, professionals use both test driven development practices and version control. These practices and their associated tools mitigate the risk for bugs in software and foster transparent and reproducible processes. Given that the tasks Tableau is used for are equally complex, it is dangerous to use processes that lack these safe guards.

In addition, seen from a time investment perspective, many of the skills one gains using Tableau are not transferrable to other software or data processing and analytics work. It has its own UI idioms and terminology that don't always line up with statistical or programming language in other domains.

I understand that writing code is no small ask and D3.js, while amazing, is non-trivial to learn. But it is worth considering all aspects of the infrastructure and whether your investment is a long game. For us, data visualization has a been long term proposition and a modernization of the kinds of data reporting we have always done. And the investment is beginning to provide a foundation selectors and bibliographers to start developing models for collection analysis.

[1] https://web.library.wisc.edu/sp/cca/lc-classification.html
[2] https://web.library.wisc.edu/sp/cca/loan-to-volume-ratios.html#All

> On Jun 25, 2019, at 6:17 PM, Natasha Allen <[log in to unmask]> wrote:
> 
> Hi folks,
> 
> Hopefully a quick question I'm doing some information gathering on. Are
> there any libraries out there currently utilizing an internal data
> dashboard for visualizing library statistics? If so, what program are you
> using for this purpose?
> 
> Thanks,
> 
> Natasha
> 
> ---
> Natasha Allen (she/her)
> System and Fulfillment Coordinator, University Library
> San José State University
> 1 Washington Square
> San José , CA 95192
> [log in to unmask]
> 408-808-2655