AI-genda

AI-Genda, The Learning Assistant

It is sometimes difficult for students to estimate how much time they should set aside for a high school subject to achieve the desired grade. To help them, this project focuses on designing a tool that gives students insight into how much they need to learn. It was a requirement to make use of machine learning, and it had to be applied in a way that only this type of technology could be used as a solution. The final design is a prototype with two parts, a software system that calculates how long you should study, and a physical part that facilitates communication between user and system. This resulted in a working prototype grounded on fabricated data that is ready to be tested with participants (high school students). The ML is a support vector regression model, which can predict the study time based on the predicted data. The SVR and two other models were tested repeatedly to see which one had the best result based on the R2 score. Using XAI, students can read the accuracy of the predicted study time, and this might help them with their planning.

Info:

Course: Embodying intelligent behavior in social context

Year: 2021

Duration: 1 Quartile

Prototype: Physical Prototype + Python Code

Coach: Kostas Tsiakas, Emilia Barakova

Team: Dillon Shieh, Jafar Fernald, Jules Sinsel, Jun Li Jeung

This course introduces me to Python and Machine Learning and that is what I primarily focused on in the team. Together with Jun Li I created the Python code used in the project to predict a students grade. While it was difficult at first when I got a hang of it, it became easy to compare different ML methods and see which one worked best with the data. I have compared over 100 different settings and methods to get to the highest results.