From: PreCount: a predictive model for correcting real-time occupancy count data
Methodology | Constraints | |
---|---|---|
Naive approach | 1. It subdivides past raw counts into daily profiles with timestamps {t0,…,tn}, where n is the end of the day. | It assumes that most buildings have periods during night time where the number of occupants go to zero. |
2. It initializes the first transition of each day i.e. transition at time t0 to zero. | ||
3. It assigns zero to all negative counts. | ||
Probabilistic approach | 1. It subdivides past raw counts into daily profiles with timestamps {t0,…,tn}, where n is the end of the day. | 1. It requires the observed maximum number of occupants to formulate and compute a transition matrix. |
2. It corrects each daily profile at time tn using specific methods proposed in either Sangoboye and Kjærgaard (2016) and Kuutti et al. (2014). | 2. It requires a complete estimation of the transition matrix propagation for each day to accurately correct obtained counts. |