trained rndc,svc,extreec using mnist dataset

Ensemble_learning/mnist_ensemble.py

@@ -14,16 +14,12 @@
 print(Y.shape)
 
 #%%
-X_train,Y_train,X_val,Y_val,X_test,Y_test = X[:40000],Y[:40000],X[40000:50000],Y[40000:50000],X[50000:],Y[50000:]
-shuffle_index = np.random.permutation(40000)
-X_train, Y_train = X_train[shuffle_index], Y_train[shuffle_index]
-X_train_array = np.array(X_train)
-Y_train_array = np.array(Y_train)
-X_val_array = np.array(X_val)
-Y_val_array = np.array(Y_val)
-X_test_array = np.array(X_test)
-Y_test_array = np.array(Y_test)
 
+from sklearn.model_selection import train_test_split
+X_train_val,X_test,Y_train_val,Y_test = train_test_split(X,Y,
+test_size=10000,random_state=42)
+X_train,X_val,Y_train,Y_val = train_test_split(X_train_val,Y_train_val,
+test_size=10000,random_state=42)
 print(X_val)
 
 #%%
@@ -33,7 +29,46 @@
 
 rnd_clf = RandomForestClassifier(n_estimators=1000,n_jobs=100,
 random_state=42)
-rnd_clf.fit(X_train_array,Y_train_array)
-rnd_pred = rnd_clf.predict(X_val_array)
-rnd_eff = accuracy_score(Y_val_array,rnd_pred)
-print(rnd_eff)
+rnd_clf.fit(X_train,Y_train)
+rnd_pred = rnd_clf.predict(X_test)
+rnd_eff = accuracy_score(Y_test,rnd_pred)
+print(rnd_eff)
+print(rnd_clf.score(X_val,Y_val))
+
+#%%
+
+from sklearn.svm import SVC
+from sklearn.preprocessing import StandardScaler
+from sklearn.pipeline import Pipeline
+svm_clf = Pipeline((
+    ("scaler",StandardScaler()),
+    ("svm_clf",SVC(kernel='rbf',gamma=5,C=0.001))
+))
+svm_clf.fit(X_train,Y_train)
+svm_pred = svm_clf.predict(X_test)
+svm_eff = accuracy_score(Y_test,svm_pred)
+print(svm_eff)
+print(svm_clf.score(X_val,Y_val))
+
+#%%
+from sklearn.ensemble import ExtraTreesClassifier
+
+extra_trees_clf = ExtraTreesClassifier(random_state=42)
+extra_trees_clf.fit(X_train,Y_train)
+extra_trees_pred = extra_trees_clf.predict(X_test)
+extra_trees_eff = accuracy_score(Y_test,extra_trees_pred)
+print(extra_trees_pred)
+print(extra_trees_clf.score(X_val,Y_val))
+
+#%%
+from sklearn.ensemble import VotingClassifier
+
+vote_clf = VotingClassifier(estimators=[
+    ('svc',svm_clf),('rnd',rnd_clf),('extra_cls',extra_trees_clf)]
+    ,voting='hard')
+
+vote_clf.fit(X_train,Y_train)
+vote_clf_predict = extra_trees_clf.predict(X_test)
+vote_clf_eff = accuracy_score(Y_test,vote_clf_predict)
+print(vote_clf_predict)
+print(vote_clf.score(X_val,Y_val))