@MASTERSTHESIS\{IMM2007-05337,
    author       = "I. Bak",
    title        = "Travel Time Forecasting",
    year         = "2007",
    school       = "Informatics and Mathematical Modelling, Technical University of Denmark, {DTU}",
    address      = "Richard Petersens Plads, Building 321, {DK-}2800 Kgs. Lyngby",
    type         = "",
    note         = "Supervised by Bo Friis Nielsen, {IMM,} {DTU,} and co-supervised by Jan Holm from the Road Directorate.",
    url          = "http://www2.compute.dtu.dk/pubdb/pubs/5337-full.html",
    abstract     = "The main objective of this thesis is the development of a forecast algorithm for short-term travel time forecasting. This algorithm is intended to become an inherent part of a new real-time traffic reporting system. This system is in the pipeline in the framework of the Danish Road Directorate. Practicability and operability are the central keywords that permeate every aspect of this thesis. Consequently great emphasis is placed on the data value chain from data collection and preparation of input data for the forecast algorithm to model deployment. The movement of data through a series of stages and the transformations that these data undergo in the process are illustrated. Data modeling is viewed as an intrinsic part of the complete data value chain with an end product in mind, combined with methods of heuristic nature. Insights into the nature of traffic data are provided by the use of clustering. The forecasting algorithm is subsequently based on the results of clustering of data. The developed algorithm is simple and its performance in terms of forecast accuracy is satisfactory. The result is considered to be superior to forecasting based on average travel times."
}