 I'm just going to skip ahead here, Martha. I'll just note on the slide here that there is another presentation next week with Rachel O'Alan, and that this is the second in the series with Last Weeks by Sarah Murphy, and on to my presentation proper here. So there are a couple of things that I'm going to talk about today. I'm going to start off with just a brief overview of the power of visualization. We'll use some data sets that I think will be familiar to many of you, the libqual and the ARL stats, as a way of understanding just what visualization can bring to our analysis of data. From that, I'm going to go on to look at Tableau specifically in the context of a broader reporting platform for UBC Library. So we've been looking at Tableau as a possibility for reporting from our Voyager ILS, and I'll report a little bit on a recent proof-of-concept project for that. And we'll leave plenty of time at the end for questions. So if you do have questions along the way, please note those and share those in the chat box as well. So because this presentation is covering a wide audience, and Tableau might be new to some of you, I thought I'd start off with just a quick message about what Tableau is, pardon me, and also what it is not. I think of Tableau as a digital data analysis and presentation layer. It's a suite of software products. So if you go to the Tableau website you see there's five or six different software offerings. But essentially this is a layer that fits on top of existing data sources. In some conversations people wonder if this is something like a data warehouse if you would be storing data in Tableau. The brief answer for that is no, that this is a tool that allows you to take existing sources and possibly even plug into databases directly and take advantage of Tableau as a way of manipulating that, of exploring that data, and of presenting it visually. I stress the visual because I think there are a couple of advantages of doing this kind of analysis visually instead of by pouring through tables of data. One is just the question of size. If you're dealing with an ILS database and multiple circulation or multiple years of circulation data you could be looking at millions of rows of data. And with visual analysis tools we can get a much quicker insight into what's actually going on in that data. And then it's also excellent to use a visual analysis tool for communicating the findings. So first for exploring and then for sharing what we find. I share this quote from Gerrith Thorpe, the former New York Times data artist in residence, fantastic job title by the way. Because it stresses that visualization and working with data in a visual way is really best thought of as a process. It's not simply about making things look pretty although that's a fortunate outcome. But I would argue that it allows us to understand our data and to process it and to ask questions of it and to engage with it in ways that we wouldn't otherwise be able to. And that's where tools like Tableau come in. Tableau is just one of many different options available for doing visual presentation and visual analysis. It's one of the tools that I have found most accessible. But it allows us to engage with data and take advantage of certain things that we know about the way we process visual information to really dig deeply and see what stories the data might hold. Next slide. I apologize this is supposed to be timed so that it could show one half first and the other half second and I think that's maybe just not possible in this presentation software. But this illustrates something that if you just look at the top half for a moment something called attentive and pre-attentive processing. So one of my favorite authors in the field of data visualization is Stephen Few and I highly recommend some of his books. These are pulled from an article shared online on his blog. Attentive processing is what's going on in the top part there. And the question that I was going to pose to you assuming the bottom half would be hidden is how many fives are in that data set. You could certainly figure that out. You could come through that bit by bit. But attentive processing is sort of a brute force just head down against the brick wall kind of processing of data. It's really what we do when we are looking at tables to some degree. But there are things that can visually make different elements pop out. This is really straightforward when you look at it, but it's a concept that I think can be applied really effectively in visual presentation. So pre-attentive processing is something that allows us to visually group things, identify things, see differences in them just at a glance. And this is a very well studied field. There are a number of different characteristics in terms of form, shape, and color that are really well suited to pre-attentive processing. And so in the second example there, it's fairly easy to see at a glance with very little cognitive load how many fives there are in the data set. I mentioned there are different types of pre-attentive processing. Even few again provide some examples. And you can see that depending on the type of data you are working with and the type of visual display, ways of just slightly changing either positioning, orientation, form, or color can make things really stand out in large data sets. And this is where a tool like Tableau comes in. It takes advantage of the research into this kind of pre-attentive processing. And it harnesses that so that you can engage with the data in this visual way without necessarily having to work through in the tabular format as I mentioned. So with that as brief introduction to some of the concepts behind visualization and some of the reasons why you could say this is worth investing in for anyone who needs to do a lot of work with large data sets. Moving on from that to some examples using Tableau. This is an ARL audience so I trust most of you are familiar with LibQual and I won't spend too much time talking about it here other than to say that when LibQual results come in you get quite a pile of numbers. And for each question there are desired, minimum, and perceived service levels for each question in different user groups. In this visual presentation here I've got an orange bar like so indicating the distance between my minimum score and my desired score. And then I've got a blue dot right there which is showing where the perceived falls within that range. And ideally I guess it does fall within that range. Sometimes it falls a bit below or above. This is all fairly straightforward but presenting this visually allows us to go beyond just a very simple presentation looking at one question only to a little bit more detail. And on the next slide shows us a few other characteristics here of the data that a tool like Tableau can allow us to extract quite readily. So what I've done here the same principle applies. I'm just looking at the columns with a bar showing minimum to desired and then a blue line in this case showing progression over time. And I have this for three LibQual periods for UBC Library. We're looking at results for the question about the printed library's materials that I need for my work. And we're looking at it in addition by broadly the subject area of the respondents. The point with bringing this slide in is that I have in one fairly compact graph quite a few layers of my LibQual data. And that if I were to look at that separately I'd be looking at a lot of numbers and trying to hold things in memory while I'm moving between one table and another. And in this case I'm able to present this all at once. And I can highlight certain things as I have here. I've highlighted that the humanities and social sciences while the general trend is for both decreasing expectations and decreasing perception of service over those years. It's also interesting to note that those fields are perhaps three to five years behind in that progression behind some of the other disciplines like the health sciences and science and math. So being able to present this visually allows me to pack a whole lot more into a small space and convey hopefully a more convincing message. And a tool like Tableau makes it possible for me to do this by simply dragging and dropping different elements into this display. It literally takes once you're familiar with the software probably under 10 minutes to go from the previous slide to something like this if your data is formatted correctly and you're somewhat familiar with what you're looking at. So another example here that's familiar to most of you from the AERAL Investment Index and this one here is just showing the possibilities with longitudinal data. I'm looking at the top quartile of public academic institutions and illustrating up here is not because I wanted to single out UCLA but because I have the opportunity when this is a live setting to hover over any of these points and get more information about the point as it is here. So this can be a way to add a lot of data to a small space. In another example, again working with the AERAL Statistics, another place where Tableau is really helpful is in calculations. It can do some of those basic calculations like medians and means and so on. It can also group libraries and you can set up automatic groupings. In this case I have a grouping setup that will show me for each year the top quartile of public academic institutions by their Investment Index ranking. And it does the calculation that takes the total library expenditures reported to AERAL and divides it by the sum of the total full-time students and the total full-time faculty to give me this approximate spending per user. We all know that there's always provises with how you use the data but if you understand what you're looking at and what your data is, this can be a really quick way of digging into the data and seeing where your library is relative to peers which can often be a message for, in this case, potentially a message for the need to increase funding. Both of these datasets have a couple things in common. They have very structured formats and they are longitudinal. And so another advantage of using Tableau is that once something like this is created you can simply update with new data and have that added on without having to redo all of your calculations. So these are two examples I've shown you so far of ways that I've used Tableau personally just in my assessment work. One of the things that I've been curious about since doing that is whether it makes sense for Ubisoft Library to expand its implementation of Tableau to use it as a reporting platform for other staff as well and for reporting needs that go beyond just those of the data analysis that I do in my job as assessment librarian. And so I'm going to talk just briefly now about a recent Ubisoft Library Tableau Proof of Concept project. Essentially we wanted to look at this against the Voyager ILS for which we have several sort of partially working reporting solutions but nothing that's completely meeting our reporting needs. And we thought we'd document Tableau's strengths and limitations in connecting to the ILS and figure out if this might be a good investment for the library to make to improve access to Ubisoft Library data. The examples I'm going to show are they're not groundbreaking and they're not especially exciting but they're meant to highlight just what Tableau can do. So bear with me we're going to jump into some circulation stats. The first table here is what we find when we connect Tableau to Voyager directly using a SQL query. So this is another powerful part of Tableau is that if you have Tableau desktop you can connect to other databases in sort of a visual way where you drag and drop tables in. But if you happen to have other reporting tools or other queries that have been used these can be repurposed in connecting directly to the database to extract exactly the dataset that you want. This kind of table you can imagine being somewhat useful at the end of a fiscal year for reporting to something as the ARL stats for example. We've also added some functionality here for other users where you can pick the fiscal year, you can pick whether you want to include the renewals and the discharges, whether you want both self-serve or staff-mediated transactions to be included. This does not take advantage of Tableau's visual layer but with that same query because I'm connecting directly to the database I can do something like what you see in this page here. This is just a partial screenshot of what we have on another page but I'll step you through what we're looking at. What it does is takes the fiscal year to date for the date that the report is run. It compares it to the fiscal year to date for the previous year and shows the difference. So in this case for our library with the largest circulation we were actually down 17% at that point in the fiscal year. That represents 40% of the library total to date. At that location just over half of our transactions were self-service. And here you can see the light line is last year by month and the dark line is this year by month lower in pretty much all of the months of the year. You see here at a glance we're pulling in a lot of detail with the exact same dataset that provided that other table but we're now starting to take advantage of data that I think frankly we underuse and of which we have much in our automated systems. It can also, seeing this kind of change can quickly point out places where a branch might be bucking the trends in the case of a few here that are actually increasing since last year or it could raise questions. For example, why does self-service account for only 22% at this branch and 58% at this one here? And I think this is a place where Tableau is really powerful as this thinking tool because it leads us to query the data in other ways and possibly to go beyond this dataset to other characteristics that help us understand what's going on in our libraries. I'll just skip on ahead to one more example. Again, this is using the exact same dataset but now for perhaps for management decisions around staffing, what I'm looking at is the hourly number of transactions per hour for an average during a selected period and then how that splits out by charges, discharges, and renewals. You'll notice here in this one is a feature we haven't seen in the others and that's an opportunity select by a branch so you can pick your branch. In the live version of this there is also a date picker so you can actually choose what date range you want whether it's just a term or whether it's several years. So we created several reports like this to test Tableau's functionality. We did some with funds and were able to show percentage left in certain funds compared to the percentage of time left in the fiscal year to see if we were on track for spending. And we documented some of the strengths and limitations. I'm going to share the strengths here and that's not because the limitations are, I'm trying to hide them but I'd say in our context the limitations were tended to be quite minor and more technical connectivity issues that we think we can resolve. Strengths in our context include alignment with campus partners. Others on campus are using Tableau so this is an opportunity for us to be not going our own way with the software and I think there's a huge benefit to integrating with campus partners on this kind of thing. Publishing and authentication options, the way we set things up at UBC Library there is potential to push this out to users through a browser and for them to authenticate using some login credentials that allow us to restrict access to reports if we want to do so. Automatic updating is also possible so reports can be set to rerun that SQL query every night for example and make sure that everything is always up to date. And interoperability is a major one, being able to connect to multiple data types. I demonstrated connecting to an Oracle database. We did the same with MySQL by connecting directly to Google Analytics and a couple of other sources as well. Of course it can connect to Excel or Access directly. I note also that if we are to go ahead with this, it's important to sink things through and make sure that you have a clear idea of what your objectives are and an owner who is accountable to this project's success and ongoing training because there is a training load and I can talk about that more during the question period if anyone wants to follow up on that. The last two slides here are about what this looks like in terms of implementation. I mentioned at the beginning that Tableau is a suite of products. What you see is roughly the workflow that goes from data source to report creation to the publishing step to the end user viewing the report. This is what the proposed workflow would look like at UBC Library. You see the different data sources there. Tableau Desktop, we anticipate having three to four people in the library who have licensed the Tableau Desktop that cost about $2,000 each. They would be the report authors who would then push the reports to something called Tableau Server which is a product UBC is licensing at a central level. And then that pushes things out to the browser. There are costs involved in this implementation particularly around Tableau Server. I don't know what that has cost UBC as a whole to implement. But being a suite of products, there are many ways you can mix and match things with Tableau. And so this is going to the opposite end of the spectrum and a very, very basic inexpensive approach would be to use just the free and public tools. And in that case you would lose the ability to connect directly to databases. So it tends to be then more for analysis of your more static data sources. Tableau Public can be used as the authoring layer but because that is free of charge there are some restrictions and you cannot save your data onto a local drive. That all has to be saved to the Tableau Public Server where of course your data is public. So maybe some restrictions or privacy regulations that limit what you can do there. If these are not barriers then the end product is virtually the same and you end up with browser-based access. It can't be controlled of course by authentication. This has been open to everyone in the world. But your same options for the end user to filter the data, to slice things in different ways, those remain. So this is actually a functional workflow in some places and I believe that a combination of this and the previous one is what Sarah Murphy was using in her presentation last week. Quick slide here if this is all new to you, where should you start? I would recommend downloading the free trial version of Tableau Desktop and browsing Tableau Public Dashboard for ideas and inspiration. And with that I'm going to stop and turn it over to questions and I have a couple of other examples of the questions lead us in that direction. So thank you. Sorry I had my muted mic and I was wondering what happened. Thank you Jeremy. This was great. Can you tell us a little bit about the training aspect you alluded to it? Tell us a little bit more about what it takes to get people trained in this environment. Sure. There's a learning curve with this software as with any. What we anticipate in implementing this, if we go ahead, and this is still an if, we did the proof of concept project but we don't have approval to go further, what I would anticipate is that there would be a handful of staff who are trained in the report creation process. That would be going from the raw data into a final product. I would say that if you're familiar with Excel and with some of Excel formulas and with pivot tables then you have the basics for understanding how Tableau deals with data. But it still would take, I would say, I'm hesitant to put a timeline on it but I just explored this on my own building on simple reports and using the online help for it. But I would say within a couple of weeks you could feel quite comfortable in working with raw data there. As far as training for end users, the way we're planning to implement this, most people in the library would just be interacting with the reports on the user side and very little training would be required. Some quick instructions on how to download the raw data or how to export his images would be, I think we'd recommend that. But for someone who's comfortable in Excel and especially who understands how pivot tables work, I think it would be quite intuitive. And I find that even after a month of casually using Tableau, it was faster than using Excel. One of the users asks that we show the last slide. So I'm going to click on that one. So while that shows, it was pointed out by Heather Scalf that there are Tableau user groups around the country and there is one in their area. Have you connected with a Tableau user group in your area? I have and I would say it's a pretty active community. So there's a Tableau user group actually here at UBC because we have several units that are actively using Tableau. And yeah, I understand there are quite a few user groups around North America and the world I imagine. So yeah, that's a great source. And the online forums I find really helpful for learning how to use the more advanced features and calculations. Great. Anything else in relation to implementation issues? Lisa Horowitz is asking about implementation issues. Yeah, I mentioned some limitations and I didn't go into them but I can talk a bit about some of the technical issues that we encountered that just may limit how we use Tableau internally. I'll qualify all of this by saying that the reason we're investigating this is that the university has made an investment in Tableau's server. And I don't know how feasible that is at other locations or whether it's something that would be accessible to a library wanting to do that independently. Because we have access to Tableau's server through that, we were exploring its possibilities during the initial period and we found that there's a bit of work that needs to happen to get the Tableau server which is outside the library communicating well with our internal systems for the automatic updates. We believe there's a firewall issue there that needs to be dealt with. There are limits on access to campus IP addresses only so I can't actually share this off campus. That's a decision taken by the university that kind of limits how we're able to use this software, a privacy consideration. I think the connectivity one for the automatic updates by connecting directly to our Voyager ILS is an important one. It highlights that if you're going to be using some of these higher end tools like Tableau's server, you will need to have IT staff who are trained in this sort of networking world and can figure out and troubleshoot some of these things as well. I hope that answers your question. And I believe you did touch upon the other question that Elizabeth Edwards is asking about the cost to UBC. He did mention the cost at the desktop level. Can you repeat that information? So the Tableau desktop is $2,000 for a license with I believe for us last year it was about $300 annual maintenance fee. And that allows you to publish to other platforms if you have access to them. At the moment we have a couple of Tableau desktop licenses. For my own work as an assessment librarian I consider this well worth the cost even if I couldn't share the reports even if it was just on my desktop. It's made it so much faster for me to navigate and work through my data sets. And I can just take a screenshot of that and share it in another setting if I need to. Tableau's server cost, I don't yet know what IT centrally for UBC would charge to the library to use that. So that's something that's still in discussion. And I can't say anything about that cost, but $2,000 would get you the desktop tool that can also connect directly to the databases. Great. And any information that you can share about Tableau online? I can't. I have no experience with it. When I started using Tableau I'm not sure it was available. My read of what I see on the website as Tableau online is something like a hosted Tableau server where it's similar services but the data is actually hosted by Tableau just not public in the way that Tableau public would be. In UBC's case I think it was important for the campus to store the data locally into a Tableau server was probably the better option. My hunch would be there would be some price variation there too but that is just a guess. You did show us this work in progress using Tableau with your ILS system. Melissa Laysam is asking whether the new ILS systems eliminate the need for Tableau. I don't know if the new ILS systems have better data visualization. Yeah. That's a really good point. I should have stressed that this proof of concept is figuring out if Tableau makes sense in our context. I would argue that at UBC Library we have quite a number of systems that have either very limited or non-existent reporting tools. And some of these are in-house databases that we've built to capture one thing or another. And access has been kind of spotty. So I think in our context even if we had an ILS that had great reporting functions because Tableau can connect to multiple data sources there would still be an argument to make that there's value there. But absolutely if you have another platform that's doing the kind of visual analysis that you need then a tool like this may in fact be redundant. I would go back to some of my earlier comments about the value of a tool like this for visual exploration. If your ILS provides reporting tools that answer most reporting questions but that still don't let you dig in and explore in a way other than through canned reports then there may still be value in having that exploration layer. But then it becomes more data visualization to understand the data than data visualization to communicate it. Great. This is the last question we'll take and maybe we can conclude with you showing us the last few examples you have. Lori Clota is thanking you and is asking, to what extent do you think that librarians and staff in the library or university staff are using the library's Tableau data for decision making? It's always a challenge I think to use data for decision making. Yeah, I can say that the handful of people with whom we've shared this within the library because it's been a fairly small project to begin with and until we are certain that we have ongoing access to Tableau Server we haven't been promoting this throughout the library as something we're promising to keep going. Those who we consulted with to build the reports have all said that this is much better than what they were using before and it has made it much easier for them to use the data in their day-to-day work. Now I qualify that by saying we talked to people who we knew already used data in their day-to-day work and so I wouldn't want to give the impression that because it's shiny and new like this that everyone will now be using data for all their decision making but I think there's a lot of promise for that. And one of the huge promises I think for increasing the likelihood that people will use the data is that if your report builders consult with your end users you can build customized reports that are really tailored to the needs of your library and that's where it might be different than an advanced ILS is it can really speak to your local needs. And I think if that's done well I can see use of library data increasing. Great. You want to show us your last example? If you would like to continue for a few more minutes over time I did prepare two more examples just in case we had the extra time so I will proceed to those. One of the things that we explored a little bit in we connected to the ILS but we thought we would also connect to other data sources and so as an example we linked to the ticket tracking or the request tracking system for IT support requests and this is I believe another Oracle database and we could pull the data in and sort things by the request type. This is a work in progress as well but the length of the bars here correlates to the number of questions that came in within a certain time period. You see here that we are just looking at requests that were created between the beginning of February and today. And then my color coding here shows how many of the requests are past the target period, how many are unresolved but still within the target period, and how many resolved within the target. What I wanted to do with this dashboard is make this a little bit more dynamic where users can actually engage with the data and set what their targets are. So in this case we have it hard coded here at 14 days to resolve, that's our target. But a user in the live version can just click into there and change that to say 7 and they can see how what our service rates are if our goal is to respond to a certain amount of requests within a 7 day period. This is I think quite helpful when you are working with data against measurable goals and objectives if you have this kind of target. And then the last slide takes advantage of this target percentage here. If I type in something like 80% in that, you see in this example here it then overlays in each of the categories and for each of the weeks reported a bar that shows you where the 80% mark is. And then you can see if the gray inches over into that 80% mark then we are still on target if that is our goal to complete 80% of our requests within two weeks. In cases where we are below that as we are in some of these past weeks here then we know that we are not quite at target. By setting these targets allowing those to be flexible people can kind of experiment and see whether a target is just unreasonable, whether there is a particular area or a particular type of request, whatever is going on in other we don't know but that might be something investigating. And a report like this I consider something I guess that you could call this a dashboard. It is not really going to answer all of your questions but it will tell you if there is an issue where you might want to look for further improvement. So just another example and another more evidence of connecting Tableau directly to another database to get very dynamic results. That is my last example Martha. Thank you. So I would like to encourage everybody to share your data visualization and analytics examples in the ARLSS at arl.org, Google group, please join the group if you are not a member of it. Thank you Jeremy for this wonderful presentation and thank you everybody for attending it.