Estimation of transit trip production using Factorial Kriging with External Drift: an aggregated data case study. Issue 4 (1st October 2016)

• Record Type: Journal Article
• Authors: Lindner, Anabele; Pitombo, Cira Souza; Rocha, Samille Santos; Quintanilha, José Alberto
• Is Part Of: Geo-spatial information science. Volume 19:Issue 4(2016)
• Page Start: 245
• Page End: 254
• Publication Date: 2016-10-01
• Subjects: Travel demand forecasting -- Geostatistics -- Kriging -- spatial statistics; Geographic information systems -- Periodicals; Cartography -- Data processing -- Periodicals; Surveying -- Data processing -- Periodicals; Remote sensing -- Periodicals; 526.0285
• DOI: 10.1080/10095020.2016.1260811
• Languages: English
• ISSNs: 1009-5020

Abstract

Abstract: Studies in transportation planning routinely use data in which location attributes are an important source of information. Thus, using spatial attributes in urban travel forecasting models seems reasonable. The main objective of this paper is to estimate transit trip production using Factorial Kriging with External Drift (FKED) through an aggregated data case study of Traffic Analysis Zones in São Paulo city, Brazil. The method consists of a sequential application of Principal Components Analysis (PCA) and Kriging with External Drift (KED). The traditional Linear Regression (LR) model was adopted with the aim of validating the proposed method. The results show that PCA summarizes and combines 23 socioeconomic variables using 4 components. The first component is introduced in KED, as secondary information, to estimate transit trip production by public transport in geographic coordinates where there is no prior knowledge of the values. Cross-validation for the FKED model presented high values of the correlation coefficient between estimated and observed values. Moreover, low error values were observed. The accuracy of the LR model was similar to FKED. However, the proposed method is able to map the transit trip production in several geographical coordinates of non-sampled values.