def make_elbow(X):
lst_k = range(1, 10)
lst_rsq = []
for k in lst_k:
kmeanModel = KMeans(n_clusters=k)
kmeanModel.fit(X)
# lst_rsq.append(np.average(np.min(cdist(X, kmeanModel.cluster_centers_, "euclidean"), axis=1)) / X.shape[0])
lst_rsq.append(
r_square(X.values, kmeanModel.cluster_centers_, kmeanModel.labels_,
k))
fig = plt.figure()
plt.plot(lst_k, lst_rsq, "bx-")
plt.xlabel("k")
plt.ylabel("RSQ score")
plt.title("The Elbow Method showing the optimal k")
plt.savefig("fig/" + type_exec + "/k-means_elbow_method")
plt.close()
plt.savefig('fig/k-means_ground_truth')
plt.close(fig)
# Compute R-square, i.e. V_inter/V
from R_square_clustering import r_square
from purity import purity_score
# Plot elbow graphs for KMeans using R square and purity scores
lst_k = range(2, 11)
lst_rsq = []
lst_purity = []
for k in lst_k:
est = KMeans(n_clusters=k)
est.fit(X_norm)
lst_rsq.append(
r_square(X_norm.to_numpy(), est.cluster_centers_, est.labels_, k))
lst_purity.append(purity_score(y.to_numpy(), est.labels_))
fig = plt.figure()
plt.plot(lst_k, lst_rsq, 'bx-')
plt.plot(lst_k, lst_purity, 'rx-')
plt.xlabel('k')
plt.ylabel('RSQ/purity score')
plt.title('The Elbow Method showing the optimal k')
plt.savefig('fig/k-means_elbow_method')
plt.close()
# # hierarchical clustering
from scipy.cluster.hierarchy import dendrogram, linkage
lst_labels = map(lambda pair: pair[0] + str(pair[1]),
cercle = plt.Circle((0, 0), 1, color='blue', fill=False)
axes.add_artist(cercle)
plt.savefig('acp_correlation_circle_axes_' + str(x_axis) + '_' +
str(y_axis))
plt.close(fig)
correlation_circle(data, 9, 2, 3)
#question 5
lst_k = range(2, 8)
lst_rsq = []
for k in lst_k:
est = KMeans(n_clusters=k)
est.fit(X_norm)
lst_rsq.append(r_square(X_norm, est.cluster_centers_, est.labels_, k))
fig = plt.figure()
plt.plot(lst_k, lst_rsq, 'bx-')
plt.xlabel('k')
plt.ylabel('RSQ')
plt.title('The Elbow Method showing the optimal k')
plt.savefig('r_square')
plt.close(fig)
est = KMeans(n_clusters=5)
est.fit(X)
# print centroids associated with several countries
lst_countries = ['EL', 'FR', 'DE', 'US']
# centroid of the entire dataset
def cah_cat():
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
print("###############################################")
print("#####RUN KMEANS DEMISSIONAIRES CATEGORIELS#####")
print("###############################################")
clients_dem = pd.read_csv('../donnees/fusion/dem.csv', sep=',')
print(clients_dem)
del clients_dem['is_adh']
del clients_dem['DTADH']
del clients_dem['DTDEM']
## Remove numerics data
del clients_dem['MTREV']
del clients_dem['AGEAD']
del clients_dem['agedem']
del clients_dem['adh']
print(clients_dem)
X_cat_one_hot = pd.get_dummies(clients_dem.astype(str))
print(X_cat_one_hot)
####
## kmeans
####
# Compute R-square, i.e. V_inter/V
# Plot elbow graphs for KMeans using R square and purity scores
lst_k = range(2, 8)
lst_rsq = []
lst_purity = []
for k in lst_k:
est = KMeans(n_clusters=k)
est.fit(X_cat_one_hot)
lst_rsq.append(
r_square(X_cat_one_hot.to_numpy(), est.cluster_centers_,
est.labels_, k))
# TODO: complete lst_purity
print("------------- Groupe de " + str(k) + " clusters ---------")
clusters = {
"code": pd.DataFrame(clients_dem.index.values.tolist()),
"cluster": est.labels_
}
print(pd.DataFrame(clusters))
fig = plt.figure()
plt.plot(lst_k, lst_rsq, 'bx-')
# plt.plot(lst_k, lst_purity, 'rx-')
plt.xlabel('k')
plt.ylabel('RSQ/purity score')
plt.title('The Elbow Method showing the optimal k')
plt.savefig('../fig/k-means_elbow_method')
plt.close()
print("###############################################")
print("#####END KMEANS DEMISSIONAIRES CATEGORIELS#####")
print("###############################################")
print('\tcentroid: ' + str(est.cluster_centers_[num_cluster]))
print(
'--------------------------------------------------------------------------'
)
print(
'--------------------------------------------------------------------------'
)
lst_k = range(2, 9)
lst_rsq = []
for k in lst_k:
est = KMeans(n_clusters=k)
est.fit(eurofit)
lst_rsq.append(r_square(eurofit, est.cluster_centers_, est.labels_, k))
fig = plt.figure()
plt.plot(lst_k, lst_rsq, 'bx-')
plt.xlabel('k')
plt.ylabel('RSQ')
plt.title('The Elbow Method showing the optimal k')
plt.savefig('R2')
plt.show()
plt.close()
print(
'--------------------------------------------------------------------------'
)
print(
'--------------------------------------------------------------------------'
)