Using Machine Learning in order to Estimate the Traffic Mix in a Building from the Stops Data

Previous work has established that the average number of up stops and down stops in a building during a round trip, as well as the ratio between them, could be used to estimate the mix of traffic prevailing in the building and its intensity. Further work has used basic correlation methods to derive the mix of traffic in the building, finding the ratios of incoming traffic, outgoing traffic, interfloor traffic. These studies have assumed that inter-entrance traffic is zero. This paper builds on the methodologies developed in the earlier work by introducing machine learning techniques to model the relationship between stop types and their locations within a building. The methodology requires knowledge of the types of floors in the building (occupant floors or entrance/exit floors). The data required for machine learning will be generated in larger amounts in a reasonable time and with modest processing power, whereby the data is representative of a specific building.