More accurate historical data for studyspace occupancy

[hueyy]

Right now, the chart in the app is plotted based on the average occupancy across the last 30 days. Whilst this generally works fine, the data can be very inaccurate during certain periods of time:

• Between holidays and term time: overestimation when the holiday begins, and underestimation for the first few weeks of term
• During bank holidays / public holidays / special events
• When a new studyspace is opened / closed / etc, e.g. when the UCL Student Centre was opened, when a studyspace is re-opened after renovation, etc.
• Certain libraries during certain periods of the year when there are many deadlines in close proximity

A better approach may be to identify analogous periods in time. The following app, which provides estimates of crossing times for one of the busiest land borders in the world, does this:

One easy way could be to look at the equivalent month/week from the previous year.

[hueyy]

Maybe a weighted average of the last 4 weeks (more heavily weighted towards more recent data than older data)