Blue collar laborers’ travel pattern recognition: Machine learning classifier approach

Abstract

This paper proposes a pattern recognition model to develop clusters of homogenous activities for blue-collar workers in the State of Qatar. The activity-based data from the travel diary of 1051 blue-collar workers collected by the Ministry of Transportation and Communication (MoTC) in Qatar was used for analysis. A pattern recognition model is applied to a revealed preference (RP) survey obtained from the Ministry of Transportation and Communication (MoTC) in Qatar for the travel diary for blue-collar workers. Raw data preprocessing and outliers detection and filtering algorithms were applied at the first stage of the analysis, and consequently, an activity-based travel matrix was developed for each household. The research methodology undertaken in this paper comprises a combination of different machine learning techniques, predominantly by applying clustering and classification methods. A bagged Clustering algorithm was employed to identify the number of clusters, then the C-Means algorithm and the Pamk algorithm were implemented to validate the results. Meanwhile, the interdependencies between the resulted clusters and the socio-demographic attributes for the households were examined using crosstabulation analysis. The study results show significant diversity amongst the clusters in terms of trip purpose, modal split, destination choice, and occupation. Furthermore, whilst the Bagged Clusters and Pamk Clusters techniques on the three attributes yielded similar results, the Cmeans Clusters differed significantly in a number of the clusters. Applying such pattern recognition models on big and complex activity datasets could assist transport planners to understand the travel needs of segments of the population well and formulating better-informed strategies.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.