查看文章

Often in practice, the problem of unavailability of specific desired knowledge within one (“target”) dataset arises. However, if this knowledge can be extracted from a different (“source”) dataset and transferred between the datasets, this could increase the value of the target dataset at relatively minimal cost. The goal of this paper is to evaluate approaches to informing one dataset with knowledge from another and to evaluate the performance of the knowledge transferred into the target dataset. We use the 2009 National Household Travel Survey as the target dataset. The missing knowledge is transportation-related attitudes, whose inclusion could greatly improve travel behavior models. Our source dataset is obtained from the 2011–12 Multitasking Survey of Northern California Commuters. To achieve the goal, the set of common variables was first augmented with a large number of built-environment attributes. Then, after applying machinelearning methods, pro-transit, pro-active transportation, and pro-density attitudinal factor scores were predicted with the greatest precision; correlations of the predicted and observed scores were 0.564, 0.538, and 0.571, respectively. The performance of the transferred attitudes was measured by estimating linear regression models of vehicle ownership. The results showed that in the source dataset the observed attitudes account for an 8.0% model lift (improvement in goodness of fit), while in the target dataset the predicted attitudes account for a 1.2–5.4% model lift. Although these initial results are modest, we believe they show substantial promise, and the process has identified a number of opportunities for …