To reduce the confusion, I changed my application from traffic to meeting room, so this application is about modeling a meeting room efficiency , the data collection is built by placing a kind of motion sensor at the center of room for detecting and counting the motion activities. So my data are time-series based for the motion activities.
The model I want to build is for the purpose of optimizing the time length for showing the people the status of room being occupied or not. If a sensor receives motion event at the moment, in order to decide how long to show the occupied status, the application would look back the history of motion events, (say last 5 minutes)
In case if counting 20 consecutive motion events in the last 5 minutes, then it means there was high density of activities, so I would assume the room was busy and will be busy in the next consecutive minutes, then make the occupied status time stay longer time (5 minutes) before it turns to free status
If in cases that in the last 5 minutes, only 2 events being monitored, then it means the room might be not busy, then make this occupied status time stay only 1 minute.
So far I'm only concerned about what kind of statistical model should I use to predict the how long the occupied time length based on the latest 5 minutes'motion history?