Analytics and Visualizations

Analytics

The predictive model displays WAP users in the selected building at each time based on the median of WAP users over the past three weeks from the selected date.

Why Median?

During the pandemic, Wifi data witnessed an unusual amount of users on certain days. The irregularity in the number of users was caused by the short periods of lockdowns placed by RPI for the students' safety. The number of users was low during a lockdown, access to buildings was restricted/limited. The numbers were high immediately after the lockdown was lifted. The median method effectively eliminates the unusual amount of users by considering the most relevant data: the number of users on a normal day.

Below are three bar plots comparing the predictive mean and predictive median of someday in April 2021 with the actual population of that building throughout the day. The absolute difference between median user count and actual user count for a time range works out to be less than the absolute difference between mean user count and the actual for that time range. Thus, the median represents the predictions with more accuracy.

Visualization

The data is displayed as a histogram that displays the number of WAP users recorded in that building every hour on that day. The selected time is highlighted on the histogram with a vertical line and a slightly darker bar.