Visualizing Confused Student EEG Brainwave Data with Seaborn

The dataset we'll be working with in this lesson is dubbed the Confused student EEG brainwave data and is available on Kaggle. It was uploaded by Haohan Wang and used within the Using EEG to Improve Massive Open Online Courses Feedback Interaction research paper by Haohan Wang et al. at Carnegie Mellon University. The aim of their study was to see if we can detect confusion from EEG data or not. If yes - this could mean that MOOC platforms could start supplying students with home EEG devices to wear during lectures. The teachers would then have solid feedback on whether their lectures are confusing or not! Teaching online, teachers have significantly less feedback on the state of the classroom, which can lead to a worse lecturing experience. If nothing else - we can test whether we can figure out if students are confused or not as reliably as with eyesight.

Naturally, pattern-recognition approaches are used to come up with the conclusion - most notably, Machine Learning algorithms are used to find intricate relationships between the data. Also naturally, exploratory data analysis is the first step in this! It's worth noting that Haohan Wang clearly states that binary classification on this dataset "extremely challenging". In essence, this means that it's extremely challenging to really predict the state of confusion based on the EEG data. This doesn't need to mean that EEG itself is not the technology to use - the equipment and processing algorithms play a huge role in this too! A major thing to note is that this was, after all, a single-channel EEG device, which limits the accuracy of the results. Because of this, we probably won't be able to see high correlation between features and the state of student confusion. If there were high, clear cut correlation, it wouldn't be very challenging to perform binary classification on the dataset.