Mining GPS Data for Trajectory Recommendation
read more
Citations
Analysis of human mobility patterns from GPS trajectories and contextual information
Review of Wearable Device Technology and Its Applications to the Mining Industry
Tour recommendation and trip planning using location-based social media: a survey
Detecting Anomalous Trajectories and Behavior Patterns Using Hierarchical Clustering from Taxi GPS Data
Big Trajectory Data: A Survey of Applications and Services
References
Maximum likelihood from incomplete data via the EM algorithm
LIBSVM: A library for support vector machines
Item-based collaborative filtering recommendation algorithms
Fab: content-based, collaborative recommendation
Related Papers (5)
Frequently Asked Questions (7)
Q2. What is the objective function for the trajectories?
Trajectory recommendation requires to give a ranking score to each candidate trajectory while classification algorithms, e.g., decision tree, can only output binary values.
Q3. What is the function for determining the value of pi?
For ABR, the ranking function is thus defined as:Scoreabr(T, u) = logn ∑i=1pi · 1T.activity=pi pi = ni ∑nj=1 nj (4)Due to the wide use of GPS-equipped smart phones, much attention is focused on the use of the trajectory data to improve people’s life, among which transportation mode detection is most related to their work.
Q4. What is the frequency of the trajectories?
the frequency of trajectory5 http://en.wikipedia.org/wiki/Window function#Overlapping windows1 lies mainly in higher spectrum while that of trajectory 2 in lower part.
Q5. How many users are removed from Everytrail?
users whose shared trajectories are less than 20 are removed since small sample may hurt the model accuracy and thus the recommendation performance.
Q6. What is the simplest way to obtain a GPS trace?
The GPS trace is obtained via GPS sensor which sampled the moving object’s current location together with the sampled time stamp.
Q7. Why is the data collected by Everytrail so noisy?
Because the trajectory data is uploaded by different people and there is no strict examination, the raw data contains much noise for mining.