csv_cluster.py

import argparse
from matplotlib import pyplot

from dengraph import dengraph
from dengraph.graphs import graph_io

CLI = argparse.ArgumentParser()
CLI.add_argument('--csv', help='csv inout file')
CLI.add_argument('-d', '--distance', help='maximum distance between nodes', default=1, type=float)
CLI.add_argument('-n', '--neighbours', help='neighbours required for cores', default=5, type=float)

options = CLI.parse_args()

with open(options.csv) as csv_file:
    graph = graph_io.csv_graph_reader((ln for ln in csv_file if ln[0] != '#' and ln != '\n'), nodes_header=True, symmetric=True)

#pyplot.hist([dist for node in graph for dist in graph[node].values()], bins=200)
if False:
    xs, ys = [], []
    for node in graph:
        for r in range(100):
            dst = r/100
            xs.append(dst)
            ys.append(len(list(graph.get_neighbours(node, dst))))

    pyplot.hist2d(xs, ys, bins=(100, 250))
    pyplot.show()

clustering = dengraph.DenGraphIO(graph, core_neighbours=options.neighbours, cluster_distance=options.distance)

for idx, cluster in enumerate(clustering):
    print(idx, len(cluster))