Meta-analytic framework for efficiently identifying progression groups in highway condition analysis
The minimum message length two-dimensional segmenter (MML2DS) criterion is a powerful technique for road condition data analysis developed at the Nottingham Transportation Engineering Centre (NTEC), University of Nottingham. The criterion analyses condition data sets by simultaneously identifying op...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Published: |
American Society of Civil Engineers
2016
|
| Online Access: | https://eprints.nottingham.ac.uk/44715/ |
| _version_ | 1848796981778972672 |
|---|---|
| author | Prince, Rawle Byrne, Matthew Parry, Tony |
| author_facet | Prince, Rawle Byrne, Matthew Parry, Tony |
| author_sort | Prince, Rawle |
| building | Nottingham Research Data Repository |
| collection | Online Access |
| description | The minimum message length two-dimensional segmenter (MML2DS) criterion is a powerful technique for road condition data analysis developed at the Nottingham Transportation Engineering Centre (NTEC), University of Nottingham. The criterion analyses condition data sets by simultaneously identifying optimum trends in condition progression, the position in time and space of maintenance interventions, longitudinal segments within links, and the error likelihood of each measurement. This is done in an unsupervised manner through classification and regression models on the basis of the minimum message length (MML) metric. Use of MML, however, often requires an exhaustive comparison of all possible models, which naturally raises considerable search-control issues. This is precisely the case with the MML2DS approach. This paper presents an efficient meta-analytic framework for controlling the generation of progression groups, which considerably reduces the search space before the application of MML2DS. This is achieved by identifying founder sets of longitudinal segments, around which families of segments are likely to be formed. An effective subset of these families is then selected, after which the MML2DS criterion is used as the final arbiter to determine ultimate model configurations and fits. This approach has proved to be very powerful, resulting in significant improvements in efficiency to the effect that accurate results are obtained in a few minutes where it previously took weeks with much smaller data sets. The indications are that this approach can be applied to other techniques besides MML2DS. |
| first_indexed | 2025-11-14T19:56:37Z |
| format | Article |
| id | nottingham-44715 |
| institution | University of Nottingham Malaysia Campus |
| institution_category | Local University |
| last_indexed | 2025-11-14T19:56:37Z |
| publishDate | 2016 |
| publisher | American Society of Civil Engineers |
| recordtype | eprints |
| repository_type | Digital Repository |
| spelling | nottingham-447152020-05-04T20:02:49Z https://eprints.nottingham.ac.uk/44715/ Meta-analytic framework for efficiently identifying progression groups in highway condition analysis Prince, Rawle Byrne, Matthew Parry, Tony The minimum message length two-dimensional segmenter (MML2DS) criterion is a powerful technique for road condition data analysis developed at the Nottingham Transportation Engineering Centre (NTEC), University of Nottingham. The criterion analyses condition data sets by simultaneously identifying optimum trends in condition progression, the position in time and space of maintenance interventions, longitudinal segments within links, and the error likelihood of each measurement. This is done in an unsupervised manner through classification and regression models on the basis of the minimum message length (MML) metric. Use of MML, however, often requires an exhaustive comparison of all possible models, which naturally raises considerable search-control issues. This is precisely the case with the MML2DS approach. This paper presents an efficient meta-analytic framework for controlling the generation of progression groups, which considerably reduces the search space before the application of MML2DS. This is achieved by identifying founder sets of longitudinal segments, around which families of segments are likely to be formed. An effective subset of these families is then selected, after which the MML2DS criterion is used as the final arbiter to determine ultimate model configurations and fits. This approach has proved to be very powerful, resulting in significant improvements in efficiency to the effect that accurate results are obtained in a few minutes where it previously took weeks with much smaller data sets. The indications are that this approach can be applied to other techniques besides MML2DS. American Society of Civil Engineers 2016-05 Article PeerReviewed Prince, Rawle, Byrne, Matthew and Parry, Tony (2016) Meta-analytic framework for efficiently identifying progression groups in highway condition analysis. Journal of Computing in Civil Engineering, 30 (3). 04015044. ISSN 1943-5487 http://ascelibrary.org/doi/10.1061/%28ASCE%29CP.1943-5487.0000518 doi:10.1061/(ASCE)CP.1943-5487.0000518 doi:10.1061/(ASCE)CP.1943-5487.0000518 |
| spellingShingle | Prince, Rawle Byrne, Matthew Parry, Tony Meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| title | Meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| title_full | Meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| title_fullStr | Meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| title_full_unstemmed | Meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| title_short | Meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| title_sort | meta-analytic framework for efficiently identifying progression groups in highway condition analysis |
| url | https://eprints.nottingham.ac.uk/44715/ https://eprints.nottingham.ac.uk/44715/ https://eprints.nottingham.ac.uk/44715/ |