 Hi everyone, I'm Sonia Trivedi, Coms Manager here at Moodle and we are at the second half of day number one of Moodle Mood Global 2023. I have really interesting guests with me today, Stefan Orenna, Chief Data and AI Officer at Essex Business School. Hi Stefan, how are you doing today? Hi Sonia, I'm fine, thanks. How is the Moodle and Moodle Mood Global here so far? Very good, it's always a pleasure to be here. Perfect, thank you for that. And you had really interesting presentation just a few minutes ago, named using Moodle Analytics to predict student success. Could you please tell me a little bit more about that? I understood you have really interesting use case about Moodle Analytics, so if you can elaborate, that would be great. Well we used Moodle Analytics, which is a brick that is inside Moodle, and we just have to, it's a user friendly interface that you can use in order to choose some machine learning models in order to evaluate or predict students at risk of failing in their online course. And so in a few clicks, you can deploy those algorithms and they make predictions and they give you a list of students at risk for a given course. And then you can contact those students and after that you will reduce the risk for them to fail. Really interesting. So I'm curious to know a little bit more insights on the project, like how did you implement it over what period of time if you can tell us more? Well we spent a few weeks trying to evaluate the solution with a consultant that we hired in order to develop some specific metrics for machine learning, so precision and recall and confusion metrics in order to see what was inside the platform because when you use the platform you only see the result of a statistical metric which name is F1 score and we wanted to see more, so we developed extra metrics in a few weeks and saw the performances of the algorithms as we wanted, let's say one month because we were working remotely with a consultant and then when we were satisfied by the results we launched on new courses directly this feature in order to predict students at risk and trying to limit this number. Okay, good. I want to ask you a little bit more about the results of the project but before we get there, can you tell our audience a little bit more about ESSEC Business School, like you implemented the project, how many courses, how many students, that kind of things? ESSEC Business School is a top business school in Europe, one of the most renowned and so we launched that project on four courses last year which involved more than 1000 students and we are planning to run it for our online courses this year. Okay, and you mentioned some figures during your presentation, can you share some of them here, like what were the results, how many students were impacted by this? Among, we had more than 1000 students involved in the course, we had a few hundreds that were at risk and so we were able to contact them and try to reduce the number of students that were failing. We are carrying out right now some A-B tests in order to measure the statistical significance of what we are carrying out right now, so we'll be able to give some figures I hope by the end of the year. Okay, so no figures for now. Well, I have figures for the machine learning algorithms, so the precision was around 82 and the recall 76, so that means that we were able to detect more than 75% of the students that are really going to fail and we are able to contact more than 80% of students that are really going to fail, so less than 20% of false alerts. So it's something that is quite robust and valuable because you don't have a lot of false alerts and at the same time you don't miss a lot of students that are really failing. Impressive, impressive. We will look forward to know more about your results and more about the things you achieve from this project and last but not least, I want to ask you how important is predictive analytics for forming data-driven pathological strategies in modern education? Well, it's really important when you have to address thousands of students, a teacher cannot address all of them like in a classroom, it's not possible, you cannot follow them especially for courses that last weeks or months. You cannot know exactly what they are doing, what is their pace, etc. So it is fundamental to deploy those methods today in order to have not only just a modern approach but in order to have real impact on your students, make sure that they succeed and it's something that is decisive in this perspective. Thank you Stefan, I wish you a great time here at MoodoMood 2023 and see you soon. Thank you. Thank you very much.