Moving Type Detection without Time Information

Today location technologies are integrated into many devices enabling location-based services. Movement data recorded with these devices can be uploaded to web sites and shared with others. Movement data can be organized using keywords and semantic tags, e.g. walking and running. Our main goal is to...

Full description

Saved in:
Bibliographic Details
Main Authors: Garbe, M., Bunnig, C., Gutschmidt, A., Cap, C.
Format: Conference Proceeding
Language:English
Subjects:
Accuracy
angle-based feature
classification algorithms
Data mining
Decision trees
Global Positioning System
GPS
Legged locomotion
Support vector machines
Tracking
trajectory
Transportation
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Today location technologies are integrated into many devices enabling location-based services. Movement data recorded with these devices can be uploaded to web sites and shared with others. Movement data can be organized using keywords and semantic tags, e.g. walking and running. Our main goal is to automatically classify movement data as walking, cycling or driving. In contrast to other work we use real world GPS data without time information. Users delete time information to protect their privacy. Over a period of two months we collected movement data from a popular track sharing platform and classified the GPS tracks using Decision Tree, Naive Bayes, Support Vector Machine and Naive Bayes Tree. Our results show that users can expect reasonable results from tag suggestion services detecting moving type even without time information. A second result of our work indicates that angle-based features are insignificant for classification using real world GPS tracks. Distance was the only significant feature in our study.
DOI:10.1109/ICSC.2012.40