Initial commit

Divers/tutoriel31/README.md

@@ -0,0 +1,5 @@
+# Tutoriel 31
+## K-moyennes: coefficient silhouette
+
+La vidéo de ce tutoriel est disponible à l'adresse suivante: https://www.youtube.com/watch?v=H_AW_lwvdDk
+

Divers/tutoriel31/silouhette.py

@@ -0,0 +1,42 @@
+import numpy as np
+from matplotlib import pyplot as plt
+from matplotlib.figure import Figure
+from matplotlib.backends.backend_agg import FigureCanvas
+from sklearn.cluster import KMeans
+from sklearn.datasets.samples_generator import make_blobs
+import cv2
+
+cluster_std=1.30
+n_samples=300
+X, y=make_blobs(n_samples=n_samples, centers=5, cluster_std=cluster_std)
+
+fig, (ax1, ax2)=plt.subplots(1, 2)
+canvas=FigureCanvas(fig)
+fig.set_size_inches(11, 6)
+
+k=2
+while 1:
+    ax1.cla()
+    ax1.scatter(X[:,0], X[:,1], marker='+', c="#FF0000")
+
+    kmeans=KMeans(n_clusters=k)
+    pred_y=kmeans.fit_predict(X)
+
+    ax2.cla()
+    ax2.scatter(X[:,0], X[:,1], c=pred_y, marker='+')
+    ax2.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s=50, c='#0000FF')
+    canvas.draw()
+
+    img=np.array(canvas.renderer.buffer_rgba())
+    cv2.putText(img, "Nbr cluster={:02d}    [p|m] nbr clusters   [r] reset  [q] quit".format(k), (250, 50), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 255), 2)
+
+    cv2.imshow("plot", img)
+    key=cv2.waitKey()&0xFF
+    if key==ord('p'):
+        k=min(99, k+1)
+    if key==ord('m'):
+        k=max(2, k-1)
+    if key==ord('r'):
+        X, y=make_blobs(n_samples=n_samples, centers=5, cluster_std=cluster_std)
+    if key==ord('q'):
+        quit()

Divers/tutoriel31/silouhette2.py

@@ -0,0 +1,43 @@
+import numpy as np
+from matplotlib import pyplot as plt
+from matplotlib.figure import Figure
+from matplotlib.backends.backend_agg import FigureCanvas
+from sklearn.cluster import KMeans
+from sklearn.datasets.samples_generator import make_blobs
+import cv2
+import glob
+
+k=5
+cluster_std=1.30
+n_samples=300
+X, y=make_blobs(n_samples=n_samples, centers=k, cluster_std=cluster_std)
+
+fig, (ax1, ax2)=plt.subplots(1, 2)
+canvas=FigureCanvas(fig)
+fig.set_size_inches(10, 6)
+
+while 1:
+    ax1.cla()
+    ax1.scatter(X[:,0], X[:,1], marker='+', c="#FF0000")
+    ax1.set_title('Données')
+
+    wcss=[]
+    for i in range(1, 11):
+        kmeans=KMeans(n_clusters=i)
+        kmeans.fit(X)
+        wcss.append(kmeans.inertia_)
+
+    ax2.cla()
+    ax2.plot(range(1, 11), wcss, c="#FF0000")
+    ax2.set_title('WCSS pour "elbow method"')
+
+    canvas.draw()
+    img=np.array(canvas.renderer.buffer_rgba())
+    cv2.putText(img, "[r] reset  [q] quit".format(k), (450, 40), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 255), 2)
+
+    cv2.imshow("plot", img)
+    key=cv2.waitKey()&0xFF
+    if key==ord('r'):
+        X, y=make_blobs(n_samples=n_samples, centers=k, cluster_std=cluster_std)
+    if key==ord('q'):
+        quit()

Divers/tutoriel31/silouhette3.py

@@ -0,0 +1,57 @@
+import numpy as np
+from matplotlib import pyplot as plt
+from matplotlib.figure import Figure
+from matplotlib.backends.backend_agg import FigureCanvas
+from sklearn.cluster import KMeans
+from sklearn.datasets.samples_generator import make_blobs
+from sklearn.metrics import silhouette_score
+import cv2
+import glob
+
+k=5
+cluster_std=1.30
+n_samples=300
+fig, ((ax1, ax2), (ax3, ax4))=plt.subplots(2, 2)
+canvas=FigureCanvas(fig)
+fig.set_size_inches(12, 8)
+
+X, y=make_blobs(n_samples=n_samples, centers=k, cluster_std=cluster_std)
+
+while 1:
+    ax1.cla()
+    ax1.plot(X[:,0], X[:,1], "+", c="#FF0000")
+    ax1.set_title('Données')
+
+    wcss=[]
+    tab_silhouette=[]
+    for i in range(2, 11):
+        kmeans=KMeans(n_clusters=i)
+        cluster_labels=kmeans.fit_predict(X)
+        wcss.append(kmeans.inertia_)
+        tab_silhouette.append(silhouette_score(X, cluster_labels))
+
+    ax2.cla()
+    ax2.plot(range(2, 11), wcss, c="#FF0000")
+    ax2.set_title('WCSS pour "elbow method"')
+
+    ax3.cla()
+    ax3.plot(range(2, 11), tab_silhouette, c="#FF0000")
+    ax3.set_title('Coefficient silhouette')
+
+    kmeans=KMeans(n_clusters=np.argmax(tab_silhouette)+2)
+    pred_y=kmeans.fit_predict(X)
+    ax4.cla()
+    ax4.scatter(X[:,0], X[:,1], c=pred_y, marker='+')
+    ax4.set_title('Données + centre clusters')
+    ax4.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s=100, c="#0000FF")
+
+    canvas.draw()
+    img=np.array(canvas.renderer.buffer_rgba())
+    cv2.putText(img, "[r] reset  [q] quit".format(k), (450, 40), cv2.FONT_HERSHEY_PLAIN, 1, (0, 0, 255), 2)
+
+    cv2.imshow("plot", img)
+    key=cv2.waitKey()&0xFF
+    if key==ord('r'):
+        X, y=make_blobs(n_samples=n_samples, centers=k, cluster_std=cluster_std)
+    if key==ord('q'):
+        quit()