Fairness-Aware Demand Prediction for New Mobility
Authors: An Yan, Bill Howe1079-1087
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on bike share and ride share datasets show that Fair ST can reduce inequity in demand prediction for multiple sensitive attributes (i.e. race, age, and education level), while achieving better accuracy than even state-of-the-art fairness-oblivious methods. |
| Researcher Affiliation | Academia | An Yan, Bill Howe Information School, University of Washington Seattle, WA, 98105 {yanan15, billhowe}@uw.edu |
| Pseudocode | No | The paper describes the model architecture and components but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper mentions that the data used is openly available, providing links to datasets, but it does not provide a concrete access statement or link to the source code for the methodology described in the paper. |
| Open Datasets | Yes | Mobility data. We obtained Seattle dockless bikeshare data from the Transportation Data Collaborative operated by the University of Washington. ... Ride Austin is a ride-hailing service operating in Austin, Texas. Rides data is openly available2. ... Socioeconomic data... obtained from the Simply Analytics database (Simply Analytics 2018). ... Weather features... obtained hourly weather data for Seattle and Austin from the National Centers for Environmental Information (NCEI) 3. ... Urban features. ... These urban datasets are all openly available 4. |
| Dataset Splits | No | For Seattle bikeshare, we use the data from October 2017 to August 2018 for training and the data from September to October, 2018 for testing. The training data contains 8040 temporal slices and the test data contains 1464 temporal slices. For Ride Austin, we use the data from August 2016 to February 2017 for training and the data from March to April 2017 for testing. The training data contains 5088 temporal slices and the test data contains 1056 slices. (No explicit mention of a validation set.) |
| Hardware Specification | No | The paper describes the experimental setup, but it does not provide specific hardware details (exact GPU/CPU models, processor types with speeds, memory amounts, or detailed computer specifications) used for running its experiments. |
| Software Dependencies | No | The paper mentions using “Adam optimizer” but does not provide specific software names with version numbers for libraries, frameworks, or other ancillary software components used for the experiments. |
| Experiment Setup | Yes | To implement Fair ST, we train 200 epochs for Seattle bikeshare and 350 epochs for Ride Austin using Adam optimizer with a batch size of 32. The learning rate starts at 0.005 and decays every 5,000 steps with a rate of 0.96. |