Multiplicative Cascades: How to model trip within cities
By: Marta Gonzalez
From: MIT
At: Instituto de Investigação Interdisciplinar, Anfiteatro
[2011-03-14]
Our work focuses on developing a model for trip lengths distribution in metropolitan areas, in which the current tendency is set by gravity like models, where trip lengths are fit with empirical OD (origin destination) data. While it is an extensively used method, its main drawback is that it lacks a way to include the spatial statistical variations of population density and services into the model. Depending on fit parameters for a particular region and a particular spatial scale without providing an understanding in the values of the used parameters. As we show in this paper, multiplicative cascade models can be used to generate heterogeneous distribution of populations that compare very well with the best-known resolution data as provided by LandScan. We proposes analytical expressions for trip length distributions that contain the multiplicative cascade parameters obtained from empirical density of population and allows to be adjusted to different scenarios of supply distribution. We present an extensive sensitivity analysis providing an insight in how the shape of the trip length distribution changes on different scenarios.