This repository contains the Jupyter notebooks and source codes used for reproducing the results published in our Digital Health 2019 paper:
Yue Liu, Helena Lee, Palakorn Achananuparp, Ee-Peng Lim, Tzu-Ling Cheng, and Shou-De Lin. 2019. Characterizing and Predicting Repeat Food Consumption Behavior for Just-in-Time Interventions. In Proceedings of the 9th International Conference on Digital Public Health (DPH2019). ACM, New York, NY, USA, 11-20. DOI: https://doi.org/10.1145/3357729.3357736
Please contact Liu Yue if you have any questions or problems.
The notebooks have been tested in Python 3.7 via Anaconda with the following packages:
- fpmc==0.0.0
- hpfrec==0.2.2.13
See requirements.txt for a complete list.
Download the data and extract the CSV and TSV files to the data directory.
Run the notebook 1-0-Data preparation.ipynb to perpare the datasets for the recommendation task.
Outputs: Several CSV files will be generated and stored in the data folder.
Run the notebooks 1-*.ipynb to perform data analysis of repeat and novel consumption. The notebook requires data files from previous steps in the data folder.
Outputs: Several reports will be generated and stored in the figure folder.
Run the notebook 2-*.ipynb to perform hyperparameter tuning for the recommendation models. The notebook requires data files from previous steps in the data folder.
Outputs: Several files will be generated and stored in the output/param folder.
Run the notebook 3-*.ipynb to perform the recommendation tasks. The notebook requires data files from previous steps in the data and output/param folders.
Outputs: Several files will be generated and stored in the model and output/result folder.