This week I allocated most of my time working on the prediction algorithm of the backend. We collected occupancy data of the two rooms manually on the HH 1300 wing and I constructed a decision tree algorithm that splits when classes are in session and when it is considered “peak hours” (Monday-Friday 2PM to 10PM). As of Friday night, the algorithm is outputting predictions but the predicted numbers have not reached the 80% accuracy benchmark yet. As we collect more data through the camera feed and the CV algorithm.
In terms of the prediction algorithm, we are slightly behind on schedule, but we as a team will devote more time over the weekend to enhance the prediction model, and we plan to have the prediction algorithm integrated with the backend by next weekend.