Using OpenStreetMap as a Data Source for Attractiveness in Travel Demand Models

Abstract

We present a methodology to extract points of interest (POIs) data from OpenStreetMap (OSM) for application in travel demand models. We use custom taglists to identify and assign POI elements to typical activities used in travel demand models. We then compare the extracted OSM data with official sources and point out that the OSM data quality depends on the type of POI and that it generally matches the quality of official sources. It can therefore be used in travel demand models. However, we recommend that plausibility checks should be done to ensure a certain quality. Further, we present a methodology for calculating attractiveness measures for typical activities from single POIs and national trip generation guidelines. We show that the quality of these calculated measures is good enough for them to be used in travel demand models. Using our approach, therefore, allows the quick, automated, and flexible generation of attractiveness measures for travel demand models.

Point-of-Interest (POI) Data Validation Methods: An Urban Case Study

“…In recent years, despite the lower number of articles, the field developed in several new directions, such as in more sophisticated statistical machine learning methods for quality assessment [24,25] and POI matching [26,27]. Recent studies also explore novel data sources beyond OSM, such as LBSNs and review websites [27,28], and broader applications of POI data [10,29,30]. The relevant approaches applicable to POI data sources were extracted from the surveyed literature and categorized into the six elements of data quality defined by ISO 19157:2013 [31], which are completeness, logical consistency, positional accuracy, temporal quality, thematic accuracy, and usability.…”

“…Errors can arise from commission, where excess data exists in a dataset, or omission, where data are missing from a dataset. The simplest approach for measuring completeness is to compare the number of points in D eval and in D ref [10,[33][34][35][36]. If a correspondence between points in D eval and in D ref was established, the proportion of points in D eval that are found in D ref , and vice versa, serves as a measure of completeness.…”

“…When D ref is unavailable, the intrinsic approach of analyzing community activity and growth rates over time can give an indication of completeness [10,18,21,30,35].…”

“…For example, when evaluating the category pair N 1 = "Secondary School" and N 2 = "school", the pair will be tokenized to the token sets L 1 = ["Secondary", "School"] and L 2 = ["school"]. The similarity of the sets will then be computed as stated in Equation (10). In the scenario where each pair of matched POI has multiple place types, the maximum score out of every combination will be taken as the final score.…”

“…Given the availability of geospatial information from different sources, POI data were used in a wide range of applications, from real estate valuations [5], disaggregated employment size estimation [6] and land use classification [7], to the transportation domain, with trip purpose inference [8], stop activity prediction [9] and travel demand modeling [10].…”