A Data-Driven Decision Support System for Mobile Telematics

SiamiNamini, Mohammad

A Data-Driven Decision Support System for Mobile Telematics

SiamiNamini, Mohammad

Permalink

Publication Type:: Thesis
Issue Date:: 2020

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download contents and abstractAdobe PDF (588.1 kB)

Adobe PDF

Download thesisAdobe PDF (2.68 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	SiamiNamini, Mohammad
dc.date.accessioned	2022-02-22T00:00:08Z
dc.date.available	2022-02-22T00:00:08Z
dc.date.issued	2020
dc.identifier.uri	http://hdl.handle.net/10453/154743
dc.description	University of Technology Sydney. Faculty of Engineering and Information Technology.	en_US.UTF-8
dc.description.abstract	Mobile telematics is an emerging technology that collects data on human behaviour using smartphones. All smartphones have internal sensors with the capability to record and transmit data to an external server. This technology is easy to use, the initial cost is low, and generates a massive amount of data which are noisy, complex, and uncertain. This opens many opportunities for data-driven decision making such as driving behaviour risk analysis, usage-based insurance, remote sensing, and fleet management. Traditional decision-making techniques are not able to work with this type of unstructured data. Thus, new techniques are needed based on advanced analytics to analyze mobile telematics. This research develops a big data-driven decision support system (DSS) for mobile telematics. The research relies on the capabilities of advanced analytics techniques, machine learning, and fuzzy logic. The research presents an innovative analytical system for mobile telematics which consists of four major components: 1) a data preparation component that prepares a trajectory dataset to a new and ready-for-analysis format; 2) a driving style pattern recognition that extracts hidden human patterns in mobile telematics using unsupervised learning and unlabelled data; 3) a fuzzy risk assessment is proposed to assess risk of drivers by fuzzy logic using extracted patterns by unsupervised learning; and 4) a missing data imputation component which is a novel Choquet Fuzzy Integral Vertical Bagging (CFIVB) algorithm to classify large labelled mobile telematics stream datasets. The proposed models were evaluated on two real-world mobile telematics datasets, namely an unlabelled dataset collected by a usage-based insurance company containing 500,000 journeys of 2500 drivers, and an anonymized driving behaviour dataset consisting of streaming data of 408 trips of 310 unique drivers. Various validation measures were used to evaluate the performance of the proposed models. The area under a curve (AUC) and accuracy are used to evaluate the classification algorithms and the Davis-Boulding index, the Calinski-Harabasz index, execution time, and mean square error are utilized to evaluate clustering algorithms and find the optimal number of clusters. The sensitivity analysis results show the proposed model is consistent across different variations of the model. The proposed DSS can be applied on all stream data risk assessments. Moreover, 29 unique driving styles were extracted from mobile telematics data and these patterns can be applied as labels for supervised learning modelling. In addition, performance measures depict the CFIVB algorithm performs well in this domain, and it can be applied for similar problems.	en_US.UTF-8
dc.format	Thesis (PhD)
dc.language.iso	en_US	en_US.UTF-8
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/154743/2/02whole.pdf
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	au.edu.uts.lib/ppc
dc.rights	info:eu-repo/semantics/openAccess
dc.title	A Data-Driven Decision Support System for Mobile Telematics	en_US.UTF-8
dc.type	Thesis
utslib.copyright.status	open_access	*

Abstract:

Mobile telematics is an emerging technology that collects data on human behaviour using smartphones. All smartphones have internal sensors with the capability to record and transmit data to an external server. This technology is easy to use, the initial cost is low, and generates a massive amount of data which are noisy, complex, and uncertain. This opens many opportunities for data-driven decision making such as driving behaviour risk analysis, usage-based insurance, remote sensing, and fleet management. Traditional decision-making techniques are not able to work with this type of unstructured data. Thus, new techniques are needed based on advanced analytics to analyze mobile telematics. This research develops a big data-driven decision support system (DSS) for mobile telematics. The research relies on the capabilities of advanced analytics techniques, machine learning, and fuzzy logic. The research presents an innovative analytical system for mobile telematics which consists of four major components: 1) a data preparation component that prepares a trajectory dataset to a new and ready-for-analysis format; 2) a driving style pattern recognition that extracts hidden human patterns in mobile telematics using unsupervised learning and unlabelled data; 3) a fuzzy risk assessment is proposed to assess risk of drivers by fuzzy logic using extracted patterns by unsupervised learning; and 4) a missing data imputation component which is a novel Choquet Fuzzy Integral Vertical Bagging (CFIVB) algorithm to classify large labelled mobile telematics stream datasets. The proposed models were evaluated on two real-world mobile telematics datasets, namely an unlabelled dataset collected by a usage-based insurance company containing 500,000 journeys of 2500 drivers, and an anonymized driving behaviour dataset consisting of streaming data of 408 trips of 310 unique drivers. Various validation measures were used to evaluate the performance of the proposed models. The area under a curve (AUC) and accuracy are used to evaluate the classification algorithms and the Davis-Boulding index, the Calinski-Harabasz index, execution time, and mean square error are utilized to evaluate clustering algorithms and find the optimal number of clusters. The sensitivity analysis results show the proposed model is consistent across different variations of the model. The proposed DSS can be applied on all stream data risk assessments. Moreover, 29 unique driving styles were extracted from mobile telematics data and these patterns can be applied as labels for supervised learning modelling. In addition, performance measures depict the CFIVB algorithm performs well in this domain, and it can be applied for similar problems.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/154743