Generating synthetic space-time paths using a cloning algorithm on activity behavior data

Recent decades have produced a considerable amount of discussion regarding an activity-based approach to urban travel demand modeling. An activity-based approach incorporates many of the complex and interrelated decisions of where to participate in out-of-home activities, when, for how long, and with whom to participate. This research implements an activity-based approach that sufficiently maintains many aspects of the decision-making process. This is accomplished through the cloning of representative activity patterns (RAPs).; Data mining techniques are applied to a 1993 activity diary data set of the Wasatch Front to produce RAPs. The training data set contains the time and location of activities for 1,738 people who live in Salt Lake County, Utah. From this database, 11 RAPs are generated that are necessary to seed information for the population simulation. The presented cloning algorithm begins by estimating a home-based population. Then, the algorithm clones the nearest activity pattern for each individual from the training data set that matches its RAP value. The path is then validated using a root mean square distance measure which enables researchers to quantify similarity.; The results produce a more realistic synthetic population than those generated in previous research, for a continuous temporal scale is maintained for a typical working day. Thus, more accurate time scales and activity types are encompassed. The database contains space-time paths for 737,665 synthetic individuals who participate in an average of six activities and travel an average of 45 kilometers with spatial accuracy equivalent to parcel level data. The database is constituted of realistic linked trips. Since the algorithms presented here were implemented in commonly used GIS software, it could be used by local and regional planning agencies. The resulting database enables the analysis of the impact of network disruptions, with respect not only to an individual, but also to the surrounding businesses. It also allows urban planners to gain insight into prevalent urban issues such as transportation congestion and urban sprawl.