NetBayesABM

Agent-based models of plant–pollinator networks with Bayesian inference (ABC).

NetBayesABM is a Python library for simulating agent-based models (ABMs) of ecological networks, focusing on plant–pollinator interactions. It provides tools to:

• Initialize agents (plants and pollinators) with different spatial configurations.
• Construct and evolve bipartite networks dynamically.
• Define and sample prior distributions (Gamma, Exponential) for interaction parameters.
• Visualize abundances, priors, networks, and degree distributions.
• Evaluate simulated networks against empirical data using multiple metrics.

🚀 Installation

Once published on PyPI, you can install with:

pip install NetBayesABM

For development (local clone):

git clone https://github.com/galeanojav/NetBayesABM.git
cd NetBayesABM
pip install -e .

📖 Quick Example

import numpy as np
import pandas as pd
from netbayesabm.classes import Environment_plant, Environment_pol
from netbayesabm.modelling import initial_network, update_totalinks, remove_zero
from netbayesabm.visualization import plot_agents, plot_priors

# --- Define plant environment (random positions) ---
df_plants = pd.DataFrame({
    "Plant_id": [1, 2, 3],
    "Plant_sp": ["rose", "daisy", "sunflower"],
    "X": [0, 0, 0],
    "Y": [0, 0, 0],
    "Plant_sp_complete": ["Rosa sp.", "Bellis perennis", "Helianthus annuus"]
})
envp = Environment_plant(df_plants, random_position=True, xmin=0, xmax=10, ymin=0, ymax=10)

# --- Define pollinators ---
df_pols = pd.DataFrame({
    "Pol_id": [1, 2],
    "Specie": ["bee", "butterfly"],
    "x": [2.0, 8.0],
    "y": [3.0, 7.0],
    "Radius": [3.0, 3.0]
})
envpol = Environment_pol(df_pols)

# --- Build bipartite network ---
B = initial_network(df_pols['Pol_id'].tolist(), df_plants['Plant_id'].tolist())

# --- Run short simulation ---
update_totalinks(50, envpol, envp, B, xmin=0, xmax=10, ymin=0, ymax=10)
remove_zero(B)

# --- Priors ---
prior_specialist = pd.Series(np.random.gamma(2, 2, size=1000))
prior_generalist = pd.Series(np.random.gamma(2, 2, size=1000))
plot_priors(prior_specialist, prior_generalist, "example_priors")

📁 Example

A complete analysis notebook is available in:

📍 examples/example.ipynb

It includes:

• Data loading and filtering
• Network construction
• All main metrics (degree, strength, clustering, betweenness)
• Fitted models and visualizations

The example uses data (CSV files in examples/Data/).

📊 Features

• Agent and environment classes (Environment_plant, Environment_pol).
• Network initialization and evolution functions.
• Visualization utilities for abundances, priors, and degree distributions.
• Evaluation metrics (Hellinger, Jensen–Shannon, KL, Wasserstein, etc.).
• Example notebooks for a quick start.

👩‍💻 Authors

• Javier Galeano — [javier.galeano@upm.es]
• Blanca Arroyo-Correa — [blanca.arroyo@ebd.csic.es]
• Mario Castro — [marioc@iit.comillas.edu]

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.

🔗 Links • Source code on GitHub [https://github.com/galeanojav/NetBayesABM]