Developing Trip Generation and Attraction Models Using High-Frequency Proxy Data

Abstract

The attraction and generation of a trip play a crucial role in transport planning and forecasting of trips. Multiple linear regression (MLR) is the most popular method of calculating trip attractions (TA) and trip generations (TG) to produce a distribution that can be used to forecast trips with updated values of independent variables such as electricity consumption, no of households, area of the land use etc. Literature shows that the surveys require independent variables to be updated; making them expensive and time consuming. This study aims to develop an MLR model for TA and TG based on available survey data from 2013 for Western Province, Sri Lanka and to update those independent variables using High Frequency (HF) proxy data for the predicted year (2019). Data from HF proxy sources such as electricity consumption, GPS customer points, and Landsat satellite imagery data have been used to update the independent variables of TG and TA for 2019. In the initial stage of this research, data from a home visit survey conducted in 2013 and land use data from the Western Province of Sri Lanka were used to develop the MLR model for TG and TA. A correlation and a regression analysis were performed using these surveyed data. According to the study, the MLR model for TG for home-based work trips has an r2 value of 0.79 and TA for general purposes (including shops and businesses) and industrial has an r2 value of 0.74 and 0.79, respectively, indicates the strong relationship between the considered variables to predict TG and TA. The model is validated with 2013 survey data and would be helpful for real-time estimation of the TG and TA of each zone.