Christopher Winward Undergraduate Dissertation 2015/16

Predicting PC usage in Sheffield University libraries

Supervised by J.Barker

Abstract

Students at a university require adequate locations to study, but the information about where is suitable for studying is not always easily accessible. A project, ShefUniPCs , has been collecting data about free PCs at Sheffield University and displaying it online  in a user-friendly format.

The core aim of this project is to take the data collected so far, and forecast how many free PCs may be available at a certain location in the future.

This will be done by splitting the data into core features, then using various machine learning techniques (such as neural networks) to produce the forecasted data.

This can then be evaluated by comparing against real data that comes in after the predictions are made (as data is collected daily by ShefUniPCs .)

We show that reasonably accurate predictions can be made on study spaces in Sheffield University, and present suggestions how to further improve the model we create.