def doSom():
conn = pymongo_utill.getConnectionToMongoDB()
db = conn['TwitterInsert']
#users,labels,screen_names = pymongo_utill.byTimeFreq(db=db,sample=225)
users,labels,screen_names = pymongo_utill.byTimeFreq(db=db,sample=10)
conn.disconnect()
#vectorizer = WordVectorizer()
#users = vectorizer.fit_transform(users)
clusterid, celldata = somcluster(data=users, nxgrid=21, nygrid=31, niter=500)
plt.xlim((-5,25))
plt.ylim((-5,35))
"""
print(len(clusterid))
for i,v in enumerate(clusterid):
print("number:%s coordinates:%s name:%s class:%s" % (i, v, screen_names[i], labels[i]))
for i, (x,y) in enumerate(clusterid):
if labels[i] == 0:
plt.plot(x,y,'-bo')
if labels[i] == 1:
plt.plot(x,y,'-ro')
plt.show()
for i, v in enumerate(clusterid):
plt.annotate(xy=v, s=int(i/7))
"""
vor = Voronoi(clusterid)
voronoi_plot_2d(vor)
for region in vor.regions:
if not -1 in region:
polygon = [vor.vertices[i] for i in region]
plt.fill(*zip(*polygon))
plt.show()
optimalK = 0
for k in range(init_k, last_k):
km = KMeans(n_clusters=k, init='k-means++', n_init=10).fit(X)
label_predic = km.labels_
score = metrics.silhouette_score(X, label_predic, metric='cosine')
if highest_score < score:
highest_score = score
optimalK = k
return optimalK
if __name__ == '__main__':
conn = pymongo_utill.getConnectionToMongoDB()
db = conn['TwitterInsert2']
feature_vecs, labels, screen_names = pymongo_utill.byTimeFreq(db=db, sample=100)
print(len(feature_vecs))
#screen_names, labels = pymongo_utill.loadUsers(db, sample=1254)
"""
where = []
for threshold in [0,1]:
where.append(np.argwhere(labels == threshold))
centers, est, est_v = KmeansForAgeEst2(db, where, screen_names, 2)
num_fig = 0
"""
"""
for i, center in enumerate(centers):
for ctr in center:
ti = range(24)
plt.figure(num_fig)
