Visualisation of Moodle Logs with Help of Custom R Shiny App (MoodleMoot Estonia 2026)

So, welcome everybody. U as I said, my name is Tminina and I'm the very last obstacle between you and your cold beverage today tonight. And uh uh just to give you a background, uh I started using uh mood since 2015. So definitely uh I am not an expert or an uh even a developer and my background is actually in veterinary medicine. So please keep me keep that in mind. I'm a vet not a developer. Now now you might ask like so why am I here right? Actually so uh there was a thing which we created. So we had this uh online course uh and um uh I played a key role there and it was a fully online course. But the problem was that I have never developed a fully online course and this course was just to give you like a hint like what what what is it what is it to study veterary medicine and uh the target the audience was high school students but we didn't collect like any personal data only okay yeah name and email uh but uh where you came from or what you did we didn't care just register and have fun. And uh we had 160 participants and the core structure just to give you a background here. So it was like semi self-paced. So again like you did what you'd liked but it was designed to be approximately two hours of work per week and we used uh H5P. Uh so and it was de so there was seven activities and each activity had a different topic. So like one topic per week. So seven weeks and there was a forum where the veterary students answered the questions for the participants and there was a one last final quiz uh which you had to solve in order to prove the completion of the course. And in the end there was a cherry on the cake. Uh you uh if you completed the course you were invited to visit the institute and have a tool not the problem right like I said that never designed an online course. So uh the first question would be like did did it actually work right and the second course how to check it like again I had no experience and the logical answer is of course logs right so I don't know how what you do for procrastination if you have like a 100 million things to do and uh the deadlines are on your neck like a ton of bricks. You click through the model and you check what the developers have done and uh and uh today my suspicions were confirmed that the model was developed by many many people and uh and the we're going to go see it. So if you look in the course, click there reports and voila, there are the logs. Never used them. And I especially like this button here. Get these logs. Right? Whoever came up with this is uh was genius, right? And I want these logs. I really want these logs. Not only do I want these logs, I want more. I want get something out of these logs, but there was a problem. There's a there was a person on my way called Jason, right? And remember, I'm a vet. I don't know who Jason is. So, I just downloaded the What could go wrong, right? There is 400,000 lines. Okay, that's a quite a lot for me. And the question is like okay 400,000 lines what does what does the model even log? What what what should I do? Right? And uh the next logical step would be of course would be that okay why not use the model built-in dogs tools. Well the actually again like I said I wanted more. So again the tools which are at the moment built into the model are not what I actually actually wanted. So actually the next logical step is build your own app right and the problem is again like I have never built an app right so how difficult could it be well there's this famous saying right we do this not because it's easy but we thought it could be easy All right. And now, so the next step was Google anti-gravity, right? Anybody used anti-gravity. Okay. So, and white coding. Let's go. Right. So, I tried to find the full prompt which I used for the first time. I didn't find it. So, but it most likely it was something like this. So help me create a shiny app, right? Make it nice, right? Very uh if you have learned prompt engineering, you have to be specific, right? Right. So I was again like I was a little bit unsure about the JSON. So again like just visualize J for me just do it. uh and very important you have to give background right so it's I said okay AI it's a online course and uh because I don't know how to analyze online course so give me suggestions and look at the lo and let's look at the logs as simple as that right well not it wasn't so straightforward uh but let's look at the anti-gravity So I'm going to this might look a little bit like intimidating in the beginning. No worries. I'm going to walk you through. So most important part here agent, right? And then here is the code, right? Don't don't look at the code, see through the code. And here we check with the agent, right? So for this like okay here's an example uh that was not too convenient. So I said okay make it more convenient. I don't want to upload files there just take the files and make it happen. So I'm going to send uh you send the prompt to agent then says oh there's an R code here generates. Okay. Now the agent is not very polite. Always says okay seems like you gave me the wrong file. No matter I will manage and it checks the files which are actually there. So here is the end result. And now again the very important part about white coding is that when you start white coding is that the green parts are the words which come come into the code and the red parts are the ones which go out of the code. Right? And the most important part about wipe coding is here. You're going to see you can see the arrow. Wait for it. accept all. All right, my work is done here. Now, uh the second best thing to go is the R studio. So, let's go to our studio now. And uh there's a very important button there, run app, right? You have to click it. And voila, we have the app. So we can see the particip number of participants, the clicks, all kinds of graphs here, engagement, attrition, funnel. I would have never come up with this, but that's there's more. So like we can click things. So deadlines, anonymous names, we can anonymize. uh we can look at the for example the final test. So here is the summary of the final test stuff. How much time did they spend like and here is a correlation table and of course we want to see it in log scale right because log scale it would be very important. We want to click and see things. Uh there's also like correlation between the clicks and the score. And my special favorite density plot, right? How much time uh did the student uh student student participants uh use? But that's not all, right? We want more. We want to see we want to visualize, right? So this is like uh the uh the week by week. So it was 7 weeks plus the introduction videos and this is on a time scale and how the clicks uh accumulated uh per task per H5P task. Again really nice to see and only takes a couple of days. So does it move? So, what is the take-h home message? If you have a problem and no one else can help and if you can find it, maybe you can solve it with a custom shiny app. Okay. And the last one. So again uh would find this work would have not been possible without the tremendous help of many many people who contributed to this course and I'm just here's our team and it was mostly le led by students I just uh provided some help there thank you thank you Dharma for presentation Do we have a question? Oh, okay. Okay, more questions. Okay. Oh, I see. Yeah, thanks. Very uh very interesting and nice talk. Uh I was just curious why did you choose to do it with R and Shiny? >> What's the most logical answer? Why? Well, I didn't know I didn't know any better. I have heard about shiny, so whatever. Let's try it. How hard can it be? Fair