Exclusion of Nonspecific Geocodes and the Representation of Ridership Profiles in Survey Data
| Autor: | Dara E. Seidl, Dan Silver, Mindy Rhindress, Julia Seltzer, Farhad Siraj, Stan Hsieh |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
050210 logistics & transportation
Geographic information system Information retrieval Intersection (set theory) business.industry Computer science Mechanical Engineering 05 social sciences Data field Construct (python library) computer.software_genre 03 medical and health sciences 0302 clinical medicine Public transport 0502 economics and business Geocoding Survey data collection 030212 general & internal medicine Data mining business computer Civil and Structural Engineering |
| Zdroj: | Transportation Research Record: Journal of the Transportation Research Board. 2643:28-33 |
| ISSN: | 2169-4052 0361-1981 |
| DOI: | 10.3141/2643-04 |
| Popis: | To construct responsive travel models, transportation modelers must have accurate geocoding data; only the addresses with specific geocodes (i.e., addresses with high specificity) are retained in the data, and addresses with nonspecific geocodes (i.e., addresses with low specificity, such as at the municipality level) are excluded, replaced, or imputed. Whether this practice introduces bias in the data and leads to potential inaccuracies in models that are subsequently constructed has not been evaluated. Between December 2014 and November 2015, the Metro-North Railroad (MNR) Origin and Destination Survey collected data from Hudson Line riders, who were asked to chronicle trips from start to finish. The accuracy and specificity of the address data varied. Some respondents provided information in all possible data fields (street address or intersection, city, state, and zip code), but their addresses were geocoded to a lower level of geographic accuracy because the batch geocoder could not interpret a valid address from the combined data. The differences between MNR riders who provided specific destination addresses and MNR riders who provided nonspecific destination addresses in New York City were analyzed. The primary cause for this difference was found to be a respondent’s reasons for traveling and destination type and not a respondent’s demographic profile or specific destination location within the New York City region. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|