from _NetworKit import ClusteringCoefficient

#external imports
import math
import logging

def clustering(G, error=0.01):
	"""
		Returns approximate average local clustering coefficient
		The maximum error can be given as a parameter and determines
		the number of samples taken.

		for details see:
			Schank, Wagner: Approximating Clustering Coefficient and Transitivity
	"""
	if G.numberOfNodes() < 100:
		return ClusteringCoefficient().avgLocal(G)
	else:
		nSamples = math.ceil(math.log(10) / (error**2)) # fixed confidence of 90%
		logging.info("taking {0} samples".format(nSamples))
		return ClusteringCoefficient().approxAvgLocal(G, nSamples)