Inferring Trip Destination Purposes for Trip Records Collected through Smartphone Apps
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Rapid developments in smartphones and Global Positioning System (GPS) technology have provided a new method for travel data collection. People’s travel trajectories can be passively collected through smartphones with built-in GPS sensors and processed to infer important attributes of travel behavior such as trip destination purposes. This paper examines the feasibility of using discrete choice models and tree-based machine learning models for trip purpose inference. Using smartphone GPS trajectories and land-use data that are open-sourced for academic research, four models are developed to classify trip destination purposes into one of thirteen categories. The models include multinomial logit, mixed logit, random forest, and gradient boosting decision tree. Because the data set is significantly unbalanced with more than 50% of the trips being home or work trips, a two-stage modeling process is applied to identify home and work trips first and then subsequently to classify trips for the remaining discretionary purposes. The results show that both discrete choice models and machine learning models can achieve more than 70% prediction accuracies for home trips and work trips, but less than 30% prediction accuracies for most of the discretionary trip purposes. The discrete choice models can accurately reproduce the market share distribution of the trip purposes, but the machine learning models fail to do so. The results imply that as purely data-driven models, machine learning models may not be the best solution for trip purpose inference because they lack the theoretical background provided by microeconomics and human psychology that are essential in explaining people’s travel and activity choices.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it