Welcome to d’s documentation!
Introduction
bigraph is a high-level OO Python package which aims to provide an easy and intuitive way of interacting with bipartite networks. In essence, this package is an extension of the NetworkX package (see here)
The aim here was to define an easy-to-use package which would allow developers to perform link pridiction, evaluation, and visualization of the results.
The current implementation has been developed and tested in Python 3.6+, but should work with Python 2.7+ (maybe with minor modifications in terms of printing and error handling) and most Debian based OSs.
Motivation
As a developer in the field of machine learning, I found it pretty hard to identify a Python package which would enable one to perform link pridiction algorithms over bipartite networks.
This package is intended to provide a quick, as well as (hopefully) easy to undestand, way of dealing-with-bipartite-networks algorithm up and running, for all those out there, who, like myself, are hands-on learners and are eager to get their hands dirty from early on.
Limitations
Algorithms are quite a few, and thus have to be piled with new ones.
Evaluations used here are techniques used for unsupervised methods, however they may have to be modified.
There are more limitations for sure if you happened to encounter any issue, you are welcomed to make contributions or open and issue on github.
bigraph package
Subpackages
bigraph.evaluation package
Submodules
bigraph.evaluation.evaluation module
- bigraph.evaluation.evaluation.evaluate(graph: networkx.classes.graph.Graph, k: int = 2, method: str = 'all')
Evaluation interface for evaluating algorithms
- Parameters
graph – Networkx bipartite graph
k – Number of folds (used in KFold)
method – Algorithm name
- Returns
Calculated metrics: overall_precision, overall_auc, fpr_algo, tpr_algo
Module contents
bigraph.preprocessing package
Submodules
bigraph.preprocessing.get_adjacents module
- class bigraph.preprocessing.get_adjacents.GetAdjacents
Bases:
object
bigraph.preprocessing.import_files module
- class bigraph.preprocessing.import_files.ImportFiles
Bases:
object- import_files(edge_csv: str = './inputs/neighbour_matrix.csv', label_id: str = './inputs/id_labels.csv', sep: str = ',', *args, **kwargs) dict
Import links and labels from csv files
- Parameters
edge_csv – A CSV file containing edge data on each line: 0,4,2 that are u,v,w which stand for node 1, node 2, weight respectively
label_id – A CSV file containing labels for nodes: 10,node_name which are ID,label for each node
sep – A separator that is the boundary for distinct entities
- Returns
links and label dataframes
bigraph.preprocessing.make_graph module
- class bigraph.preprocessing.make_graph.MakeGraph
Bases:
object- make_graph(dataframe: dict, left_bipartite: str = 'left_side', right_bipartite: str = 'right_side')
Make a graph based on the input dataframe
- Parameters
left_bipartite – Left part of the graph
right_bipartite – Right part of the graph
dataframe – Bipartite graph dataframe
- Returns
Generated networkx graph
bigraph.preprocessing.pd_to_list module
Module contents
- class bigraph.preprocessing.GetAdjacents
Bases:
object
- class bigraph.preprocessing.ImportFiles
Bases:
object- import_files(edge_csv: str = './inputs/neighbour_matrix.csv', label_id: str = './inputs/id_labels.csv', sep: str = ',', *args, **kwargs) dict
Import links and labels from csv files
- Parameters
edge_csv – A CSV file containing edge data on each line: 0,4,2 that are u,v,w which stand for node 1, node 2, weight respectively
label_id – A CSV file containing labels for nodes: 10,node_name which are ID,label for each node
sep – A separator that is the boundary for distinct entities
- Returns
links and label dataframes
- class bigraph.preprocessing.MakeGraph
Bases:
object- make_graph(dataframe: dict, left_bipartite: str = 'left_side', right_bipartite: str = 'right_side')
Make a graph based on the input dataframe
- Parameters
left_bipartite – Left part of the graph
right_bipartite – Right part of the graph
dataframe – Bipartite graph dataframe
- Returns
Generated networkx graph
Submodules
bigraph.algorithms module
- class bigraph.algorithms.Algorithms
Bases:
object- static adamic_adar(set_one: list, set_two: list, graph) float
Calculate Adamic Adar score for input lists
- Parameters
set_one – A list of graph nodes -> part one
set_two – A list of graph nodes -> part two
graph – NetworkX bipartite graph
- Returns
Adamic Adar score
- static common_neighbors(set_one: list, set_two: list) int
Calculate Common neighbors score for input lists
- Parameters
set_one – A list of graph nodes -> part one
set_two – A list of graph nodes -> part two
- Returns
Common neighbours score
- static jaccard(set_one: list, set_two: list) float
Calculate Jaccard score for input lists
- Parameters
set_one – A list of graph nodes -> part one
set_two – A list of graph nodes -> part two
- Returns
Jaccard score
- static katz_similarity(node_i: int, node_j: int, graph) float
Calculate Katz score for input nodes
- Parameters
node_i – Starting node
node_j – Destination node
graph – NetworkX bipartite graph
- Returns
Katz similarity score
- static preferential_attachment(set_one: list, set_two: list) int
Calculate Preferential attachment score for input lists
- Parameters
set_one – A list of graph nodes -> part one
set_two – A list of graph nodes -> part two
- Returns
Preferential attachment score
bigraph.bigraph module
- class bigraph.bigraph.BiGraph
Bases:
bigraph.algorithms.Algorithms,bigraph.preprocessing.import_files.ImportFiles,bigraph.preprocessing.make_graph.MakeGraph,bigraph.preprocessing.get_adjacents.GetAdjacents- aa_predict() dict
Compute the Jaccard-Needham dissimilarity between two 1-D arrays.
- Returns
A dictionary containing the Adamic-adar score for left_element and right_element.
- cn_predict() dict
Return the common neighbors of two nodes in a graph.
- Returns
A dictionary containing the Common neighbours score for left_element and right_element.
- jc_predict() dict
Compute the Jaccard-Needham dissimilarity between two 1-D arrays.
- Returns
A dictionary containing the Jaccard distance between vectors left_element and right_element.
- katz_predict(df_nodes: dict) dict
Compute the Katz similarity score of all node pairs.
- Parameters
df_nodes – Graph nodes
- Returns
A dictionary containing the Preferential attachment score for left_element and right_element.
- pa_predict() dict
Compute the preferential attachment score of all node pairs.
- Returns
A dictionary containing the Preferential attachment score for left_element and right_element.
Module contents
A package for link prediction in bipartite networks.
- class bigraph.BiGraph
Bases:
bigraph.algorithms.Algorithms,bigraph.preprocessing.import_files.ImportFiles,bigraph.preprocessing.make_graph.MakeGraph,bigraph.preprocessing.get_adjacents.GetAdjacents- aa_predict() dict
Compute the Jaccard-Needham dissimilarity between two 1-D arrays.
- Returns
A dictionary containing the Adamic-adar score for left_element and right_element.
- cn_predict() dict
Return the common neighbors of two nodes in a graph.
- Returns
A dictionary containing the Common neighbours score for left_element and right_element.
- jc_predict() dict
Compute the Jaccard-Needham dissimilarity between two 1-D arrays.
- Returns
A dictionary containing the Jaccard distance between vectors left_element and right_element.
- katz_predict(df_nodes: dict) dict
Compute the Katz similarity score of all node pairs.
- Parameters
df_nodes – Graph nodes
- Returns
A dictionary containing the Preferential attachment score for left_element and right_element.
- pa_predict() dict
Compute the preferential attachment score of all node pairs.
- Returns
A dictionary containing the Preferential attachment score for left_element and right_element.