01095765

Component

This paper describes a heuristic imputation algorithm that generates "cleaned" detector data with a full timestamp grid. The technique uses a simple error diagnostics algorithm by focusing on missing data, spikes, and repeating zeros, and it detects the majority of the “bad” data that is identified by more sophisticated criteria. The imputation process combines temporal interpolation and historical averages by differentiating short term breaks and long term missing data. Instead of selecting the raw archived data for historical averages, a moving average method that incorporates previously cleaned data is used to generate new estimations. The algorithm is implemented by the Traffic Operations and Safety (TOPS) Laboratory at the University of Wisconsin-Madison to support detector data analysis at the Wisconsin DOT Statewide Traffic Operations Center and Regional offices. It efficiently generates off-line imputations of detector data every 24 hours for 1,467 freeway operations detectors in southeast and north central Wisconsin.

01084478

08-2583

English

17p

2008

Transportation Research Board 87th Annual Meeting

DVD

Figures (5) ; References (10) ; Tables (7)

Advanced traffic management systems; Algorithms; Data quality; Error analysis; Heuristic methods; Information management; Traffic data; Traffic estimation; Traffic models

Wisconsin Department of Natural Resources

Data and Information Technology; Highways; Operations and Traffic Management; Planning and Forecasting; I10: Economics and Administration

Transportation Research Board Annual Meeting 2008 Paper #08-2583

BTRIS, TRIS, TRB

Jan 29 2008 5:00PM