Business Problem and Background: We are going to found a company turning the redundant space in campus into appealing co- working spaces for students. Since we have no any history record for predicting future demand of discussion room. The existing booking data of discussion room of library can, to some extent, reflect the pattern of real needs. Therefore, the project is going to forecast the demand of discussion room as accurate as possible to help our company better prepare and allocate resources.

Forecasting Goal: Predict future weekly booking in the following semester

Implement Detail and Results: We have only two semesters data available to predict. We identify that the main users are from college of engineering (CE) and technology management (CTM). Therefore, we build two models to predict each of them. After capturing the pattern of two colleges. We try to further predict potential customer component of the CTM, hence, we build three models to predict three main departments of CTM, finding out that only prediction of department quantitative finance is relative reliable.

Recommendations and Limitations: Based on the results, we would suggest that we have to prepare the maximum of capacity of seat is about 200 (combining two colleges results), and we would suggest the regularly maintenance should be performed on the week 63(detail showing in the figure in report). However, the goal of using history record to customize some activities and promotion campaigns for specific department may have some concerns. Our suggestion is to wait for more data collecting and do roll-forward forecasting. And I also suggest the future work should devote to building one reliable model to do all forecasting work to reduce labor