2020
Dane, G.; Feng, T.; Luub, F.; Arentze, T.
Route choice decisions of E-bike users: Analysis of GPS tracking data in the Netherlands Book
2020, ISSN: 18632351.
Abstract | Links | BibTeX | Tags: Big data, E-bike, GPS, Route choice
@book{Dane2020b,
title = {Route choice decisions of E-bike users: Analysis of GPS tracking data in the Netherlands},
author = {G. Dane and T. Feng and F. Luub and T. Arentze},
doi = {10.1007/978-3-030-14745-7_7},
issn = {18632351},
year = {2020},
date = {2020-01-01},
journal = {Lecture Notes in Geoinformation and Cartography},
abstract = {Over the past years, the usage of electric bikes has emerged. E-bikes are suitable for short and medium distance trips. Therefore, the Dutch government promotes using e-bikes for daily commuting trips. However, the impact of increasing demand on the cycling infrastructure is unclear. Additionally, route choice models for e-bikes are limited. This paper estimates a route choice model for e-bike users in the Noord-Brabant region of The Netherlands. The data used are based on 17626 trips from 742 users including user profiles extracted from GPS data. In order to analyze the data, a mixed logit model is applied on the route choice of respondents with addition of the path-size attribute. Mixed logit model allows a panel data setup and enables the examination of preference heterogeneity around the mean of distance attribute. Moreover, the path-size attribute is included on the model to account for the overlap between alternatives. Socio-demographic characteristics and trip-related factors are found to be influencing on the route choice decisions of e-bike and bike users. There are differences on the significance of variables between e-bike and bike users.},
keywords = {Big data, E-bike, GPS, Route choice},
pubstate = {published},
tppubtype = {book}
}
Over the past years, the usage of electric bikes has emerged. E-bikes are suitable for short and medium distance trips. Therefore, the Dutch government promotes using e-bikes for daily commuting trips. However, the impact of increasing demand on the cycling infrastructure is unclear. Additionally, route choice models for e-bikes are limited. This paper estimates a route choice model for e-bike users in the Noord-Brabant region of The Netherlands. The data used are based on 17626 trips from 742 users including user profiles extracted from GPS data. In order to analyze the data, a mixed logit model is applied on the route choice of respondents with addition of the path-size attribute. Mixed logit model allows a panel data setup and enables the examination of preference heterogeneity around the mean of distance attribute. Moreover, the path-size attribute is included on the model to account for the overlap between alternatives. Socio-demographic characteristics and trip-related factors are found to be influencing on the route choice decisions of e-bike and bike users. There are differences on the significance of variables between e-bike and bike users.
2019
Feng, T.; Timmermans, H. J. P.
Integrated imputation of activity-travel diaries incorporating the measurement of uncertainty Journal Article
In: Transportation Planning and Technology, vol. 42, iss. 3, 2019, ISSN: 10290354.
Abstract | Links | BibTeX | Tags: Activity-travel diary, GPS, imputation, travel survey, trip patterns, uncertainty
@article{Feng2019,
title = {Integrated imputation of activity-travel diaries incorporating the measurement of uncertainty},
author = {T. Feng and H. J. P. Timmermans},
doi = {10.1080/03081060.2019.1576384},
issn = {10290354},
year = {2019},
date = {2019-01-01},
journal = {Transportation Planning and Technology},
volume = {42},
issue = {3},
abstract = {Procedures to transform GPS tracks into activity-travel diaries have been increasingly addressed due to their potential benefit to replace traditional methods used in travel surveys. Existing approaches for data annotation however are not sufficiently accurate, which normally involves a prompted recall survey for data validation. Imputation algorithms for transportation mode detection seem to be largely dependent on speed-related features, which may blur the quality of classification results, especially with transportation modes having similar speeds. Therefore, in this paper we propose an enhanced integrated imputation approach by incorporating the critical indicators related to trip patterns, reflecting the effects of uncertain travel environments, including bus stops and speed percentiles. A two-step procedure which embeds a segmentation model and a transportation mode inference model is designed and examined based on purified prompted recall data collected in a large-scale travel survey. Results show the superior performance of the proposed approach, where the overall accuracy at trip level reaches 93.2% and 88.1% for training and surveyed data, respectively.},
keywords = {Activity-travel diary, GPS, imputation, travel survey, trip patterns, uncertainty},
pubstate = {published},
tppubtype = {article}
}
Procedures to transform GPS tracks into activity-travel diaries have been increasingly addressed due to their potential benefit to replace traditional methods used in travel surveys. Existing approaches for data annotation however are not sufficiently accurate, which normally involves a prompted recall survey for data validation. Imputation algorithms for transportation mode detection seem to be largely dependent on speed-related features, which may blur the quality of classification results, especially with transportation modes having similar speeds. Therefore, in this paper we propose an enhanced integrated imputation approach by incorporating the critical indicators related to trip patterns, reflecting the effects of uncertain travel environments, including bus stops and speed percentiles. A two-step procedure which embeds a segmentation model and a transportation mode inference model is designed and examined based on purified prompted recall data collected in a large-scale travel survey. Results show the superior performance of the proposed approach, where the overall accuracy at trip level reaches 93.2% and 88.1% for training and surveyed data, respectively.
Dane, G.; Borgers, A.; Feng, T.
Subjective immediate experiences during large-scale cultural events in cities: A geotagging experiment Journal Article
In: Sustainability (Switzerland), vol. 11, iss. 20, 2019, ISSN: 20711050.
Abstract | Links | BibTeX | Tags: Cultural event, Event visitors, Geotagging, GPS, Subjective immediate experiences
@article{Dane2019,
title = {Subjective immediate experiences during large-scale cultural events in cities: A geotagging experiment},
author = {G. Dane and A. Borgers and T. Feng},
doi = {10.3390/su11205698},
issn = {20711050},
year = {2019},
date = {2019-01-01},
journal = {Sustainability (Switzerland)},
volume = {11},
issue = {20},
abstract = {Cities are increasingly exploiting new activities such as large-scale cultural events in public open spaces. Investigating the subjective immediate experiences of visitors is valuable to reflect on these events and their configuration in the city. Therefore the aim of this study is twofold: (i) to demonstrate a data collection methodology to measure subjective immediate experiences of visitors and (ii) to test different types of factors that influence visitors' subjective immediate experiences at cultural events by means of the new methodology. A quantitative research that is enabled by geotagging, paper surveys and secondary data (location characteristics and weather conditions) is applied at the Dutch Design Week event in Eindhoven, the Netherlands. After data collection, a binary logit model is estimated. It is found that apart from age and intended duration of stay, visitor characteristics do not influence the subjective immediate experiences while temporal, physical environmental and weather conditions do. Specifically, it is found that subjective immediate experiences at outdoor locations are mainly influenced by location characteristics. This study shows that the proposed data collection methodology is useful for gathering insights especially on the influence of physical characteristics on subjective immediate experiences. The paper concludes with recommendations for future research and with suggestions to policy makers and event managers.},
keywords = {Cultural event, Event visitors, Geotagging, GPS, Subjective immediate experiences},
pubstate = {published},
tppubtype = {article}
}
Cities are increasingly exploiting new activities such as large-scale cultural events in public open spaces. Investigating the subjective immediate experiences of visitors is valuable to reflect on these events and their configuration in the city. Therefore the aim of this study is twofold: (i) to demonstrate a data collection methodology to measure subjective immediate experiences of visitors and (ii) to test different types of factors that influence visitors’ subjective immediate experiences at cultural events by means of the new methodology. A quantitative research that is enabled by geotagging, paper surveys and secondary data (location characteristics and weather conditions) is applied at the Dutch Design Week event in Eindhoven, the Netherlands. After data collection, a binary logit model is estimated. It is found that apart from age and intended duration of stay, visitor characteristics do not influence the subjective immediate experiences while temporal, physical environmental and weather conditions do. Specifically, it is found that subjective immediate experiences at outdoor locations are mainly influenced by location characteristics. This study shows that the proposed data collection methodology is useful for gathering insights especially on the influence of physical characteristics on subjective immediate experiences. The paper concludes with recommendations for future research and with suggestions to policy makers and event managers.
2015
Feng, T.; Timmermans, H. J. P.
Detecting activity type from gps traces using spatial and temporal information Journal Article
In: European Journal of Transport and Infrastructure Research, vol. 15, iss. 4, 2015, ISSN: 15677141.
Abstract | Links | BibTeX | Tags: Activity type, GPS, Random forest
@article{Feng2015,
title = {Detecting activity type from gps traces using spatial and temporal information},
author = {T. Feng and H. J. P. Timmermans},
url = {https://journals.open.tudelft.nl/ejtir/article/view/3103},
doi = {10.18757/ejtir.2015.15.4.3103},
issn = {15677141},
year = {2015},
date = {2015-01-01},
urldate = {2015-01-01},
journal = {European Journal of Transport and Infrastructure Research},
volume = {15},
issue = {4},
abstract = {Detecting activity types from GPS traces has been important topic in travel surveys. Compared to inferring transport mode, existing methods are still relatively inaccurate in detecting activity types due to the simplicity of their assumptions and/or lack of background information. To reduce this gap, this paper reports the results of an endeavour to infer activity type by incorporating both spatial information and aggregated temporal information. Three machine learning algorithms, Bayesian belief network, decision tree and random forest, are used to investigate the performance of these approaches in detecting activity types. The test is based on GPS traces and prompted recall data, collected in the Rijnmond region, The Netherlands. Results show that the random forest model has the highest accuracy. The model incorporating spatial and temporal information can predict activity types with an accuracy of 96.8% for the used dataset. These findings are expected to benefit research on the use of GPS technology to collect activity-travel diary data.},
keywords = {Activity type, GPS, Random forest},
pubstate = {published},
tppubtype = {article}
}
Detecting activity types from GPS traces has been important topic in travel surveys. Compared to inferring transport mode, existing methods are still relatively inaccurate in detecting activity types due to the simplicity of their assumptions and/or lack of background information. To reduce this gap, this paper reports the results of an endeavour to infer activity type by incorporating both spatial information and aggregated temporal information. Three machine learning algorithms, Bayesian belief network, decision tree and random forest, are used to investigate the performance of these approaches in detecting activity types. The test is based on GPS traces and prompted recall data, collected in the Rijnmond region, The Netherlands. Results show that the random forest model has the highest accuracy. The model incorporating spatial and temporal information can predict activity types with an accuracy of 96.8% for the used dataset. These findings are expected to benefit research on the use of GPS technology to collect activity-travel diary data.
Feng, T.; Timmermans, H. J. P.
Detecting spatial and temporal route information of GPS traces Proceedings Article
In: 2015, ISSN: 18632351.
Abstract | Links | BibTeX | Tags: Bayesian belief network, GPS, Map matching, Road network
@inproceedings{Feng2015b,
title = {Detecting spatial and temporal route information of GPS traces},
author = {T. Feng and H. J. P. Timmermans},
doi = {10.1007/978-3-319-11463-7_5},
issn = {18632351},
year = {2015},
date = {2015-01-01},
journal = {Lecture Notes in Geoinformation and Cartography},
volume = {214},
abstract = {This paper aims at detecting route information of GPS traces to represent spatial and temporal information of trips. A Bayesian belief network model is used to calculate the probability of a road matching a GPS log point. The algorithm incorporates road network topology, distance from trace nodes to road segments, the angle between two lines, direction difference, accuracy of measured GPS log point, and position of roads. GPS data collected in the Eindhoven region, The Netherlands, is used to examine the performance of this algorithm. Results based on a small sample show that the algorithm has a good performance in both processing efficiency and prediction accuracy of correctly identified instances. Prediction accuracy using a small sample is 87.02 %.},
keywords = {Bayesian belief network, GPS, Map matching, Road network},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper aims at detecting route information of GPS traces to represent spatial and temporal information of trips. A Bayesian belief network model is used to calculate the probability of a road matching a GPS log point. The algorithm incorporates road network topology, distance from trace nodes to road segments, the angle between two lines, direction difference, accuracy of measured GPS log point, and position of roads. GPS data collected in the Eindhoven region, The Netherlands, is used to examine the performance of this algorithm. Results based on a small sample show that the algorithm has a good performance in both processing efficiency and prediction accuracy of correctly identified instances. Prediction accuracy using a small sample is 87.02 %.
Feng, Tao; Timmermans, Harry J P
Detecting spatial and temporal route information of GPS traces Proceedings Article
In: Ivan, Igor; Benenson, Itzhak; Jiang, Bin; Horák, Jiri; Haworth, James; Inspektor, Tomás (Ed.): pp. 61-75, Kluwer Academic Publishers, 2015, ISBN: 9783319114620, (11th Symposium on Geoinformatics for Intelligent Transportation (GIS Ostrava 2014), GIS Ostrava 2014 ; Conference date: 27-01-2014 Through 29-01-2014).
Abstract | Links | BibTeX | Tags: Bayesian belief network, GPS, Map matching, Road network
@inproceedings{Feng2015d,
title = {Detecting spatial and temporal route information of GPS traces},
author = {Tao Feng and Harry J P Timmermans},
editor = {Igor Ivan and Itzhak Benenson and Bin Jiang and Jiri Horák and James Haworth and Tomás Inspektor},
doi = {10.1007/978-3-319-11463-7_5},
isbn = {9783319114620},
year = {2015},
date = {2015-01-01},
journal = {Geoinformatics for Intelligent Transportation},
pages = {61-75},
publisher = {Kluwer Academic Publishers},
abstract = {This paper aims at detecting route information of GPS traces to represent spatial and temporal information of trips. A Bayesian belief network model is used to calculate the probability of a road matching a GPS log point. The algorithm incorporates road network topology, distance from trace nodes to road segments, the angle between two lines, direction difference, accuracy of measured GPS log point, and position of roads. GPS data collected in the Eindhoven region, The Netherlands, is used to examine the performance of this algorithm. Results based on a small sample show that the algorithm has a good performance in both processing efficiency and prediction accuracy of correctly identified instances. Prediction accuracy using a small sample is 87.02 %.},
note = {11th Symposium on Geoinformatics for Intelligent Transportation (GIS Ostrava 2014), GIS Ostrava 2014 ; Conference date: 27-01-2014 Through 29-01-2014},
keywords = {Bayesian belief network, GPS, Map matching, Road network},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper aims at detecting route information of GPS traces to represent spatial and temporal information of trips. A Bayesian belief network model is used to calculate the probability of a road matching a GPS log point. The algorithm incorporates road network topology, distance from trace nodes to road segments, the angle between two lines, direction difference, accuracy of measured GPS log point, and position of roads. GPS data collected in the Eindhoven region, The Netherlands, is used to examine the performance of this algorithm. Results based on a small sample show that the algorithm has a good performance in both processing efficiency and prediction accuracy of correctly identified instances. Prediction accuracy using a small sample is 87.02 %.
2013
Feng, T.; Timmermans, H. J. P.
Transportation mode recognition using GPS and accelerometer data Journal Article
In: Transportation Research Part C: Emerging Technologies, vol. 37, 2013, ISSN: 0968090X.
Abstract | Links | BibTeX | Tags: Accelerometer, Activity type, Bayesian belief network, GPS, Transportation mode
@article{Feng2013b,
title = {Transportation mode recognition using GPS and accelerometer data},
author = {T. Feng and H. J. P. Timmermans},
doi = {10.1016/j.trc.2013.09.014},
issn = {0968090X},
year = {2013},
date = {2013-01-01},
journal = {Transportation Research Part C: Emerging Technologies},
volume = {37},
abstract = {Potential advantages of global positioning systems (GPS) in collecting travel behavior data have been discussed in several publications and evidenced in many recent studies. Most applications depend on GPS information only. However, transportation mode detection that relies only on GPS information may be erroneous due to variance in device performance and settings, and the environment in which measurements are made. Accelerometers, being used mainly for identifying peoples' physical activities, may offer new opportunities as these devices record data independent of exterior contexts. The purpose of this paper is therefore to examine the merits of employing accelerometer data in combination with GPS data in transportation mode identification. Three approaches (GPS data only, accelerometer data only and a combination of both accelerometer and GPS data) are examined. A Bayesian Belief Network model is used to infer transportation modes and activity episodes simultaneously. Results show that the use of accelerometer data can make a substantial contribution to successful imputation of transportation mode. The accelerometer only approach outperforms the GPS only approach in terms of the predictive accuracy. The approach which combines GPS and accelerometer data yields the best performance. © 2013 Elsevier Ltd.},
keywords = {Accelerometer, Activity type, Bayesian belief network, GPS, Transportation mode},
pubstate = {published},
tppubtype = {article}
}
Potential advantages of global positioning systems (GPS) in collecting travel behavior data have been discussed in several publications and evidenced in many recent studies. Most applications depend on GPS information only. However, transportation mode detection that relies only on GPS information may be erroneous due to variance in device performance and settings, and the environment in which measurements are made. Accelerometers, being used mainly for identifying peoples’ physical activities, may offer new opportunities as these devices record data independent of exterior contexts. The purpose of this paper is therefore to examine the merits of employing accelerometer data in combination with GPS data in transportation mode identification. Three approaches (GPS data only, accelerometer data only and a combination of both accelerometer and GPS data) are examined. A Bayesian Belief Network model is used to infer transportation modes and activity episodes simultaneously. Results show that the use of accelerometer data can make a substantial contribution to successful imputation of transportation mode. The accelerometer only approach outperforms the GPS only approach in terms of the predictive accuracy. The approach which combines GPS and accelerometer data yields the best performance. © 2013 Elsevier Ltd.
2011
Feng, T; Arentze, T A; Timmermans, H J P
Instantaneous emission modeling with GPS-based vehicle activity data: results of diesel trucks for one-day trips Journal Article
In: Journal of the Eastern Asia Society for Transportation Studies, vol. 9, pp. 756-771, 2011, ISSN: 1881-1124.
Abstract | Links | BibTeX | Tags: Emission, GPS, Transport environment
@article{Feng2011b,
title = {Instantaneous emission modeling with GPS-based vehicle activity data: results of diesel trucks for one-day trips},
author = {T Feng and T A Arentze and H J P Timmermans},
url = {https://www.jstage.jst.go.jp/article/eastpro/2011/0/2011_0_147/_article},
doi = {https://doi.org/10.11175/eastpro.2011.0.147.0},
issn = {1881-1124},
year = {2011},
date = {2011-01-01},
urldate = {2011-01-01},
journal = {Journal of the Eastern Asia Society for Transportation Studies},
volume = {9},
pages = {756-771},
abstract = {This paper presents an instantaneous analysis for traffic emissions using GPS-based vehicle activity data. The different driving conditions, including real-time and average speed, short-time stops and long-time stops, acceleration and deceleration, etc., are extracted from GPS data. The hot emission, cold-start emission and idling emission, varied by nitrogen compounds and particulate matter are calculated, respectively, in terms of the driving condition and vehicle characteristics. Results simulated based on a one-day trip activity dataset show that trucks spend most kilometers on national roads, followed by municipal and provincial roads. The number of short-time stops is significantly higher than long-time stops, and the time spent for long-time stops is higher than short-time duration. The hot emission accounts for the largest proportion of emissions, and the idling emission also contribute substantially. Results of sensitivity analyses indicate that pollutions in urban area from freight transport can be significantly decreased by increasing the vehicle classes and guiding the heavy trucks out of the region.},
keywords = {Emission, GPS, Transport environment},
pubstate = {published},
tppubtype = {article}
}
This paper presents an instantaneous analysis for traffic emissions using GPS-based vehicle activity data. The different driving conditions, including real-time and average speed, short-time stops and long-time stops, acceleration and deceleration, etc., are extracted from GPS data. The hot emission, cold-start emission and idling emission, varied by nitrogen compounds and particulate matter are calculated, respectively, in terms of the driving condition and vehicle characteristics. Results simulated based on a one-day trip activity dataset show that trucks spend most kilometers on national roads, followed by municipal and provincial roads. The number of short-time stops is significantly higher than long-time stops, and the time spent for long-time stops is higher than short-time duration. The hot emission accounts for the largest proportion of emissions, and the idling emission also contribute substantially. Results of sensitivity analyses indicate that pollutions in urban area from freight transport can be significantly decreased by increasing the vehicle classes and guiding the heavy trucks out of the region.
Feng, T; Moiseeva, A; Timmermans, H J P
Processing of national travel survey GPS pilot data : a technical report prepared for the Department for Transport Technical Report
2011.
Abstract | Links | BibTeX | Tags: GPS
@techreport{Feng2011,
title = {Processing of national travel survey GPS pilot data : a technical report prepared for the Department for Transport},
author = {T Feng and A Moiseeva and H J P Timmermans},
url = {https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/230562/Processing_of_NTS_GPS_Pilot_Data_a_technical_report.pdf},
year = {2011},
date = {2011-01-01},
urldate = {2011-01-01},
publisher = {NTS Publications},
abstract = {1.1 In November 2010, the National Centre for Social Research (NatCen)
contracted Eindhoven University of Technology to undertake GPS data processing
for a pilot of the National Travel Survey (NTS) for Great Britain, which used
accelerometer equipped Global Positioning System (GPS) devices to collect
personal travel data to replace the paper travel diary.
1.2 This report presents the background, technical details and application results
of the data processing stage of the NTS GPS pilot project. It documents the
development and application of a tool, called TraceAnnotator - developed by the
team to process (semi-)automatically multi-day GPS traces - which was then applied
to the data collected by NatCen. During this work we further improved our algorithms
by increasing complexity and identifying specific conditions, or even by visual
inspection and manual correction.
1.3 Details on the data collection and further background to this pilot project
conducted for the Department for Transport (Great Britain) can be found in the
National Travel Survey 2011 GPS Pilot Field Report, by Josi Rofique, Alun
Humphrey and Caroline Killpack (NatCen, August 2011)1.
1.4 The key requirements of the tasks described in this report were to:
Input into designing new questions for the NTS placement/pick-up
interviews to aid data processing;
Process data – including the matching of GPS data to interview data
and Geographic Information System (GIS) data, and
Technical documentation of the data processing.
1.5 We were required to clean and process the data into trip and trip stages and
the infer mode and purpose of journey. Outputs were also to include the journey start
and end point and the length of the journey (distance and time).
1.6 The GPS data were collected for 874 respondents aged 12 or more during the
seven day travel week that followed the NTS pilot survey, alongside additional
information collected during the CAPI placement and pick-up interviews to assist
data processing. Data was collected using the MGEData Mobitest GSL
accelerometer- equipped GPS device2. Questions were added to the standard NTS
interviews on personal points of interest (schools, work, gym, supermarkets etc),details of when respondents failed to charge or carry their devices, and where
atypical, the respondent’s hours of work.
1.7 A number of amendments were made to the existing TraceAnnotator system
to process the NTS GPS data. Development work was done to use the
accelerometer traces to infer transport modes and link the resulting algorithm to the
TraceAnnotator system. Some key variables had to be estimated as devices were
not set-up correctly; requiring further adjustment and the nature of some of the
training data also required additional adjustments.},
keywords = {GPS},
pubstate = {published},
tppubtype = {techreport}
}
1.1 In November 2010, the National Centre for Social Research (NatCen)
contracted Eindhoven University of Technology to undertake GPS data processing
for a pilot of the National Travel Survey (NTS) for Great Britain, which used
accelerometer equipped Global Positioning System (GPS) devices to collect
personal travel data to replace the paper travel diary.
1.2 This report presents the background, technical details and application results
of the data processing stage of the NTS GPS pilot project. It documents the
development and application of a tool, called TraceAnnotator – developed by the
team to process (semi-)automatically multi-day GPS traces – which was then applied
to the data collected by NatCen. During this work we further improved our algorithms
by increasing complexity and identifying specific conditions, or even by visual
inspection and manual correction.
1.3 Details on the data collection and further background to this pilot project
conducted for the Department for Transport (Great Britain) can be found in the
National Travel Survey 2011 GPS Pilot Field Report, by Josi Rofique, Alun
Humphrey and Caroline Killpack (NatCen, August 2011)1.
1.4 The key requirements of the tasks described in this report were to:
Input into designing new questions for the NTS placement/pick-up
interviews to aid data processing;
Process data – including the matching of GPS data to interview data
and Geographic Information System (GIS) data, and
Technical documentation of the data processing.
1.5 We were required to clean and process the data into trip and trip stages and
the infer mode and purpose of journey. Outputs were also to include the journey start
and end point and the length of the journey (distance and time).
1.6 The GPS data were collected for 874 respondents aged 12 or more during the
seven day travel week that followed the NTS pilot survey, alongside additional
information collected during the CAPI placement and pick-up interviews to assist
data processing. Data was collected using the MGEData Mobitest GSL
accelerometer- equipped GPS device2. Questions were added to the standard NTS
interviews on personal points of interest (schools, work, gym, supermarkets etc),details of when respondents failed to charge or carry their devices, and where
atypical, the respondent’s hours of work.
1.7 A number of amendments were made to the existing TraceAnnotator system
to process the NTS GPS data. Development work was done to use the
accelerometer traces to infer transport modes and link the resulting algorithm to the
TraceAnnotator system. Some key variables had to be estimated as devices were
not set-up correctly; requiring further adjustment and the nature of some of the
training data also required additional adjustments.
contracted Eindhoven University of Technology to undertake GPS data processing
for a pilot of the National Travel Survey (NTS) for Great Britain, which used
accelerometer equipped Global Positioning System (GPS) devices to collect
personal travel data to replace the paper travel diary.
1.2 This report presents the background, technical details and application results
of the data processing stage of the NTS GPS pilot project. It documents the
development and application of a tool, called TraceAnnotator – developed by the
team to process (semi-)automatically multi-day GPS traces – which was then applied
to the data collected by NatCen. During this work we further improved our algorithms
by increasing complexity and identifying specific conditions, or even by visual
inspection and manual correction.
1.3 Details on the data collection and further background to this pilot project
conducted for the Department for Transport (Great Britain) can be found in the
National Travel Survey 2011 GPS Pilot Field Report, by Josi Rofique, Alun
Humphrey and Caroline Killpack (NatCen, August 2011)1.
1.4 The key requirements of the tasks described in this report were to:
Input into designing new questions for the NTS placement/pick-up
interviews to aid data processing;
Process data – including the matching of GPS data to interview data
and Geographic Information System (GIS) data, and
Technical documentation of the data processing.
1.5 We were required to clean and process the data into trip and trip stages and
the infer mode and purpose of journey. Outputs were also to include the journey start
and end point and the length of the journey (distance and time).
1.6 The GPS data were collected for 874 respondents aged 12 or more during the
seven day travel week that followed the NTS pilot survey, alongside additional
information collected during the CAPI placement and pick-up interviews to assist
data processing. Data was collected using the MGEData Mobitest GSL
accelerometer- equipped GPS device2. Questions were added to the standard NTS
interviews on personal points of interest (schools, work, gym, supermarkets etc),details of when respondents failed to charge or carry their devices, and where
atypical, the respondent’s hours of work.
1.7 A number of amendments were made to the existing TraceAnnotator system
to process the NTS GPS data. Development work was done to use the
accelerometer traces to infer transport modes and link the resulting algorithm to the
TraceAnnotator system. Some key variables had to be estimated as devices were
not set-up correctly; requiring further adjustment and the nature of some of the
training data also required additional adjustments.