These project files is for a Pattern Recognition course project in Mahidol University International College. This project is a Recommendation System using Collaboration Filtering.

What I attempt to do is to predict how much a person would rate a class (out of 5) based on their previous ratings of other classes and their general preferences (How much they prefer Homework, Projects, Math, Theory etc). The dataset is collected by myself through survey-based questionnaires.

Results - Even though there were some success in this project, I don't think its completed to it's full potential. Some points are as follows:

• Data is overfitted
• Must give weight/ preference to courses following a pre-requisite course. For example, if i take Pattern Recog. 1, there would be a higher chance of taking Pattern Recog. 2.
• Biases
• Forms being submitted 2+ times by the same person
• Attempting to quantize subjective feelings about a course