Data and Code for “Optimising Self-Organised Volunteer Efforts in Response to the COVID-19 Pandemic”

Citation

Please cite our paper if you find it is interesting.

@article{zhang2022optimising,
  title={Optimising self-organised volunteer efforts in response to the COVID-19 pandemic},
  author={Zhang, Anping and Zhang, Ke and Li, Wanda and Wang, Yue and Li, Yang and Zhang, Lin},
  journal={Humanities and Social Sciences Communications},
  volume={9},
  number={1},
  pages={1--12},
  year={2022},
  publisher={Palgrave}
}

NCE computation: Compute self-organisational intervals on Shenzhen’s data

Note:

1. O_NCE.csv. T_NCE.csv and P_NCE.csv are pre-computed NCEs for Shenzhen and its district using data files “issuer_task_data.csv” and “issuer_user_data” (In the data.zip file).
2. The “task label” column in organizer_task_data.csv represents the task type extracted from task descriptions using LDA.

• Label 1: Transportational Topic tasks;
• Label 2: volunteering topic tasks;
• Label 3: Reopening Topic tasks;
• Label 4: Educational topic tasks;
• Label 5: environmental topic tasks;
• Label 6: Covid-19 topic tasks.

3. “neigborhood_1.csv” and “neighborhood_2.csv” are data for case studies.

To run:

1. Run NCE.ipynb to generate an NCE plot with color shaded self-organization intervals

Causality Analysis: Causality analysis on what dynamic factors have caused self-organization events.

Note:

1. causality_data.csv contains three types NCE, internal and external variables (policies impulse and covid-19 daily new cases)
2. all_diff_data.csv is differencing from causality_data.csv to make sure our time-series data is stationary for causality analysis

To run:

1. Install tigramite package from https://github.com/jakobrunge/tigramite.git
2. Install graphviz package from https://graphviz.org/download/
3. Run Causality_analysis.ipynb to obtain full causal graphs for self-organization intervals

Simulation part: A simulation of users participating in a fixed number of tasks.

Simulation rules:

1. Simulation is initialized with a fixed number of agents and tasks
2. Each task is represented by a cell in a 2D grid. All tasks have a limit on the number of agents it can recruit: max_agent_per_cell;
3. At each step, each agent decides whether to participate in a task with probability, p_participate.
4. If the agent is participating, it will join the first available task from its recent participation history within a time window, ordered by highest frequency; (To simulate the user behavior of participating in the same task.) If no space is available for all these tasks, it will join a random available task nearby the current task.

To run:

1. Install mesa package from https://mesa.readthedocs.io/
2. Run UserModel.py to run the simulation
3. paint_subplot_simulation.py is used to draw NCEs and gains under different parameters. People can change the parameters in UserModel.py to get NCE and gains.