Requirements:
- Python 3
- Java
- Maven
- Pip
- Python packages in requirements.txt
- Online-PSL
Python package requirement are provided in requirements.txt.
pip3 install -r scripts/requirements.txt
This project requires online-PSL to run (https://github.com/linqs/online-psl). Ensure both Maven and Java are installed. Then clone the Online-PSL repository and execute "mvn clean install" at its root.
git clone https://github.com/linqs/online-psl
mvn clean install
To load the PSL formatted data used in the paper run the data/fetchData.sh
script from the data directory.
cd data
./fetchData.sh
To reconstruct the PSL formatted data from the raw datasets first modify the DATASET
variable in
scripts/data_construction/construct.py
and scripts/data_construction/predicate_constructors.py
to select the dataset you plan to run. Modify the DATA_URL
, DATA_FILE
, and EXTRACTED_DATA_FILE
to
the appropriate values for the raw dataset. Then run:
python3 scripts/data_construction/construct.py
To run the experiments, execute the script.
./run.sh
This will run all the experiments reported in the paper.
Experiment options can be modified in scrips/run_experiments.sh
If you find this repository useful please cite the following paper:
@inproceedings{dickens:hicss23,
title = {Online Collective Demand Forecasting for Bike Sharing Services},
author = {Charles Dickens and Alex Miller and Lise Getoor},
booktitle = {Hawaii International Conference on System Sciences},
year = {2023},
_publisher = {HICSS},
address = {Maui},
}