Abstract
The special events such as games, concerts, state fairs, etc. attract a large amount of population, which requires proper planning of transit services to meet the induced demand. Previous studies have proposed methods for estimating an average daily weekday demand, which have an inherent disadvantage in estimating the demand for a special event. We solve an idealized version of this problem i.e., we decompose a special event affected demand matrix into a regular and an outlier matrix. We start with detecting the special events in large scale transit data using the Mahalanobis distance, an outlier detection method for high dimensional data. Then, a special event demand is evaluated using state-of-the-art dimensionality reduction technique known as robust principal component analysis (RPCA), which is formulated as a convex optimization program. We show the application of the proposed method using Automatic Passenger Count (APC) data from Twin Cities, MN, USA. The methods are general and can be applied to any type of data related to the flow of passengers available with respect to time. Of practical interest, the methods are scalable to large-scale transit systems.
Original language | English (US) |
---|---|
Pages (from-to) | 7370-7382 |
Number of pages | 13 |
Journal | IEEE Transactions on Intelligent Transportation Systems |
Volume | 22 |
Issue number | 12 |
DOIs | |
State | Published - Dec 1 2021 |
Bibliographical note
Publisher Copyright:© 2000-2011 IEEE.
Keywords
- Mahalanobis distance
- Special event
- automatic passenger count (APC)
- origin-destination (O-D) matrix
- outlier detection
- robust principal component analysis (RPCA)
- transit data