infantnetwork

Python package for constructing networks from infant transfer records. Functions and example code is provided to construct networks and describe their shape and structure.

infantnetwork uses the computational backends NetworkX and igraph.

Code is released under the open-source MIT License.

Installation

Install the package using pip:

pip install infantnetwork

Usage

from infantnetwork import computeNetwork, sample_transfers

#Compute network using default settings
output = computeNetwork(sample_transfers)

#Extract graphs
graph_networkx = output['graph_networkx']
graph_igraph = output['graph_igraph']

#Metrics
network_metrics = output['metrics']

#Extract a single network metric
network_metrics['centrality_mean_receiving']

The network_metrics atribute of computeNetwork returns the following quantatitve network metrics:

Metric	Description
n_nodes	Number of nodes in the network.
n_edges	Total number of edges in the network.
n_transfers	Total number of transfers in the network.
n_self_loops	Number of self-loops.
centrality_median	Median node Katz centrality.
centrality_mean_receiving	Mean node Katz centrality of nodes with incoming transfers.
density_unweighted	Network density (total proportion of possible edges observed)
density_weighted	Weighted network density: $$\frac{\sum \text{(edge weights)}}{\text{(n edges)} \cdot \max(\text{(edge weights)})}$$
efficiency_global	Global network efficiency.
efficiency_median_local	Median of local node efficiencies.
modularity_greedy	Network modularity using the greedy community modularity algorithm.
modularity_randomwalk	Network modularity using random walktrap algorithm
max_node_percentage_by_component	Percentage of nodes in the largest connected component.
max_weight_percentage_by_component	Percentage of weight (transfers) within the largest connected component.