plt.scatter(x16, y16, c='purple')
plt.scatter(x17, y17, c='navy')
plt.scatter(x18, y18, c='violet')
plt.scatter(x19, y19, c='purple')
plt.scatter(x20, y20, c='coral')
plt.show()
#result2 is the predicted class of every sample
#label is the autual class of samples
label = []
result2 = []
for i in range(len(result[1])):
label.append(int(result[1][i].category))
for j in range(len(clf.labels_)):
result2.append(clf.labels_[j])
import evaluate as A
purity = A.purity(result2, label)
NMI = A.NMI(result2, label)
TP, TN, FP, FN = A.contingency_table(result2, label)
rand_index = A.rand_index(result2, label)
precision = A.precision(result2, label)
recall = A.recall(result2, label)
F_measure = A.F_measure(result2, label)
print("Purity:" + str(purity))
print("Precision:" + str(precision))
print("Recall:" + str(recall))
print("F_measue:" + str(F_measure))
if p.clusterId == i:
cluster['cluster'].add(p)
self.clusters.append(cluster)
if __name__ == '__main__':
es, ps, fs = [], [], []
for i in xrange(1, 10):
points = readPoints('dataset2.dat', 'dataset2-label.dat')
dbscan = DBscanClustering(points, i, 100)
dbscan.clustering()
es.append(i)
ps.append(purity(dbscan.clusters, dbscan.points))
fs.append(fscore(dbscan.clusters, dbscan.points))
print '\t'.join([str(i) for i in es])
print '\t'.join([str(i) for i in ps])
print '\t'.join([str(i) for i in fs])
print "\n\n"
ms, ps, fs = [], [], []
for i in xrange(10, 200, 10):
points = readPoints('dataset2.dat', 'dataset2-label.dat')
dbscan = DBscanClustering(points, 6, i)
dbscan.clustering()
d = distance(point, cluster["mean"])
ds.append(d)
minId = ds.index(min(ds))
if point.clusterId != minId:
self.clusters[point.clusterId]["cluster"].remove(point)
self.clusters[minId]["cluster"].add(point)
point.clusterId = minId
stop = False
self._reCalculateMean()
if __name__ == "__main__":
ks, ps, fs = [], [], []
for k in xrange(5, 26):
points = readPoints("dataset1.dat", "dataset1-label.dat")
kmeans = KmeansClustering(points, k)
kmeans.clustering()
ks.append(k)
ps.append(purity(kmeans.clusters, kmeans.points))
fs.append(fscore(kmeans.clusters, kmeans.points))
print "\t".join([str(i) for i in ks])
print "\t".join([str(i) for i in ps])
print "\t".join([str(i) for i in fs])