diff = 0
for i in xrange(k):
diff += distance.euclidean(newCentroids[i], oldCentroids[i])
logger.debug( 'means total diff %f:' % diff)
if diff < delta:
break
else:
oldCentroids[:] = newCentroids
np.savetxt(centroidsinputfile, newCentroids)
logger.debug( 'total time: %f' %total)
#Calculate SSE
SSE = Utils.calcSSE(points, newCentroids)
logger.info('%.f' % SSE)
logger.debug( 'Sum of Squared Error: %s' % SSE)
#per data batch
#finds the distance to nearest cluster
distmatrix = distance.cdist(points, newCentroids, metric='euclidean')
labels = distmatrix.argmin(axis=1)
#Local
#points = np.loadtxt(inputfile1)
# Plot.plotPoints(points, labels, title='final kmeans2')
# Plot.plotMeans(newCentroids)
# strnow = datetime.now().strftime("%Y-%m-%d_%H%M%S")
# plt.savefig('%s_%s.png' % (strnow,fileoutpostfix))
