def plot_figures(som):
#BMU map
vhts = BmuHitsView(12, 12, "Hits Map", text_size=12)
#U matrix
u = sompy.umatrix.UMatrixView(50,
50,
'umatrix',
show_axis=True,
text_size=8,
show_text=True)
UMAT = u.build_u_matrix(sm, distance=1, row_normalized=False)
#Cluster map
sm.cluster(6)
hits = HitMapView(10, 10, "Clustering", text_size=12)
#Show factor influence
view2D = View2D(15, 15, "time-series", text_size=10, names=names)
#Show plots
view2D.show(sm, col_sz=4, which_dim="all", denormalize=True)
vhts.show(sm,
anotate=True,
onlyzeros=False,
labelsize=12,
cmap="Greys",
logaritmic=False)
UMAT = u.show(sm,
distance2=1,
row_normalized=False,
show_data=True,
contooor=True,
blob=False)
a = hits.show(sm)
plt.show()
def som_kmeans_clustering_predict(som, k):
# This performed K-means clustering with k clusters on the SOM grid to PREDICT clusters
#[labels, km, norm_data] = som.cluster(K,K_opt)
map_labels = som.cluster(n_clusters=k)
data_labels = np.array([map_labels[int(k)] for k in som._bmu[0]])
hits = HitMapView(20, 20, "Clustering", text_size=12)
a = hits.show(som)
return som, map_labels
def plot_clusters(nb_clusters, sm):
sm.cluster(nb_clusters)
hits = HitMapView(12, 12, "Clustering", text_size=10, cmap=plt.cm.jet)
a = hits.show(sm,
anotate=True,
onlyzeros=False,
labelsize=7,
cmap="Pastel1")
plt.show()
view2D = View2D(10,10,"", text_size=7) view2D.show(sm, col_sz=5, what = 'codebook',)#which_dim="all", denormalize=True) plt.show() # Number of people in each neuron from sompy.visualization.bmuhits import BmuHitsView vhts = BmuHitsView(12,12,"Hits Map",text_size=7) vhts.show(sm, anotate=True, onlyzeros=False, labelsize=10, cmap="autumn", logaritmic=False) # Apply k- means over SOM # K-Means Clustering from sompy.visualization.hitmap import HitMapView sm.cluster(3)# <n_clusters> hits = HitMapView(10,10,"Clustering",text_size=7) a=hits.show(sm, labelsize=12) #Apply hierarchical clustering on the top of SOM #k means - hierarchical # Apply first k means with lots of centroids (huge number!!) # Then, apply hierarchical clustering over the k means centroids # SOM - K-means? # SOM: is less sensitive to outliers # SOM - hierarchical # Same as we do with k means - hierarchical
# 14.1.1.1 Silhouette scores
# Average silhouette score
silhouette_avg_som_hc_cons = silhouette_score(std_cons, som_hc_cons['Hierarchical Clustering'].values)
# Silhouette scores individual to each observation
sample_silhouette_som_hc_cons = pd.DataFrame(
silhouette_samples(std_cons, som_hc_cons['Hierarchical Clustering'].values), columns=['Value'])
# Number of positives silhouette scores
pos_sample_hc_cons = sample_silhouette_som_hc_cons[sample_silhouette_som_hc_cons.Value > 0].count()
# 14.1.2 K-Means Clustering on top of SOM
# Visualize to which of the k cluster from the k-means belongs each neuron
k = 3
som_kmeans_cons = sm_consump.cluster(k)
hits = HitMapView(10, 10, "Clustering", text_size=7)
a = hits.show(sm_consump)
# 'som_kmeans_cons' is a dataframe with a column 'K-means' that specifies to which cluster belongs each client
som_kmeans_cons = pd.DataFrame(som_kmeans_cons, columns=['K_means'])
som_kmeans_cons['Labels'] = range(mapsize_consump * mapsize_consump)
som_kmeans_cons = final_clusters_consump.merge(som_kmeans_cons, how='inner', on='Labels', right_index=True)
som_kmeans_cons = som_kmeans_cons.sort_index()
# Verify the number of observations associated of each cluster and the cluster centroids coordinates
count_obs_som_kmeans_cons = som_kmeans_cons.groupby('K_means').count()
centroids_som_kmeans_cons = som_kmeans_cons.groupby('K_means').mean()
centroids_som_kmeans_cons = centroids_som_kmeans_cons.drop(columns='Labels')
# 14.1.2.1 silhouette scores
# Average silhouette score
silhouette_avg_som_k_means_cons = silhouette_score(std_cons, som_kmeans_cons['K_means'].values)
center, radius = [0.5, 0.5], 0.5
verts = np.vstack([np.sin(theta), np.cos(theta)]).T
circle = mpath.Path(verts * radius + center)
axes2.set_boundary(circle, transform=axes2.transAxes)
plt.savefig(os.path.join(output_path, 'Map_AMET_E_ERAI.jpg'), dpi=400)
print '*******************************************************************'
print '*************** K-means clustering *****************'
print '*******************************************************************'
# determine how many groups we want to divide
# this is closely relate to physical aspect of view
# for instance, if there are 4 weather regimes in the Arctic
# then it is better to make 4 group
cl = som.cluster(n_clusters=4)
setattr(som, 'cluster_labels', [0, 1, 2, 3])
hits = HitMapView(5, 7, 'Weather regimes clustering', text_size=12)
hits.show(som)
hits.save(os.path.join(output_path, 'K_AMET_E_ERAI.jpg'), dpi=400)
print '*******************************************************************'
print '*************** U matrix visualization *****************'
print '*******************************************************************'
u = sompy.umatrix.UMatrixView(5,
7,
'U-Matrix of SLP',
show_axis=True,
text_size=12,
show_text=True)
# U matrxi value
UMat = u.build_u_matrix(som, distance=1, row_normalized=False)
# visualization
UMat = u.show(som,
distance2=1,
view2D = View2D(10, 10, "rand data", text_size=12)
view2D.show(sm, col_sz=4, which_dim="all", desnormalize=True)
# U-matrix plot
from sompy.visualization.umatrix import UMatrixView
umat = UMatrixView(width=10, height=10, title='U-matrix')
umat.show(sm)
# do the K-means clustering on the SOM grid, sweep across k = 2 to 20
from sompy.visualization.hitmap import HitMapView
K = 20 # stop at this k for SSE sweep
K_opt = 18 # optimal K already found
[labels, km, norm_data] = sm.cluster(K, K_opt)
hits = HitMapView(20, 20, "Clustering", text_size=12)
a = hits.show(sm)
import gmplot
gmap = gmplot.GoogleMapPlotter(54.2, -124.875224, 6)
j = 0
for i in km.cluster_centers_:
gmap.marker(i[0], i[1], 'red', title="Centroid " + str(j))
j += 1
gmap.draw("centroids_map.html")
from bs4 import BeautifulSoup
def insertapikey(fname, apikey):
from sompy.visualization.mapview import View2D
view2D = View2D(4,4,"rand data",text_size=16)
view2D.show(som, col_sz=2, which_dim="all", desnormalize=True)
# U-matrix plot
from sompy.visualization.umatrix import UMatrixView
umat = UMatrixView(width=10,height=10,title='U-matrix')
umat.show(som)
from sompy.visualization.hitmap import HitMapView
K=10
Kluster = som.cluster(K)
hits = HitMapView(20,20,"K-Means Clustering",text_size=16)
a=hits.show(som)
#
som.cluster(n_clusters=K)
#som.cluster() returns the k-means cluster labels for each neuron of the map,
#but it is straightforward to retrieve the cluster labels for the whole training set,
#by assigning them the label of the BMUs (best-matching units). You can can do for example:
#Make sure indices line up....
map_labels = som.cluster(n_clusters=K)
# som._bmu[0]
data_labels = np.array([map_labels[int(k)] for k in som._bmu[0]])
clusters = pd.Series(data_labels)
clusters = clusters.rename('cluster').to_frame()
#concat cluster column with small original som input df
df_labeled = None
def draw_cluster_map(self):
from sompy.visualization.hitmap import HitMapView
hits = HitMapView(20, 20, "Clustering", text_size=12)
hits.show(self.sm)
plt.show()
umat = UMatrixView(width=20, height=20, title='U-matrix')
umat.show(som)
from sompy.visualization.hitmap import HitMapView
from sompy.visualization.bmuhits import BmuHitsView
bmuhitsview = BmuHitsView(12, 12, 'Data per node', text_size=24)
bmuhitsview.show(som,
anotate=False,
onlyzeros=False,
labelsize=7,
logaritmic=False)
Kluster = som.cluster(5)
hits = HitMapView(20, 20, "K-Means Clustering", text_size=16)
a = hits.show(som, anotate=False, labelsize=7, cmap='viridis')
def HowManyK(k):
'''compute SSE for up to k clusters'''
SSE = np.empty(0)
K = np.arange(2, k)
for i in K:
totalERROR = 0
map_labels = som.cluster(
n_clusters=i) # will eventually return more than labels....
data_labels = np.array([
map_labels[int(x)] for x in som._bmu[0]
]) # mapping labels from size of grid to total size of df
clusters = pd.Series(data_labels)