Commit Live PR

q01_bagging/build.py

@@ -14,5 +14,25 @@
 
 
 # Write your code here
-
-
+def bagging(X_train, X_test, y_train, y_test, n_est):
+    dtree = DecisionTreeClassifier(random_state=9)
+    lst_train_score = list()
+    lst_test_score = list()
+    estimator_arr = range(1,n_est,2)
+    for est_val in estimator_arr:
+        bagging_clf = BaggingClassifier(base_estimator=dtree, n_estimators=est_val,\
+                                    max_samples=0.67, max_features=0.67, bootstrap=True,\
+                                    random_state=9)
+        bagging_clf.fit(X_train, y_train)
+        y_pred_test = bagging_clf.predict(X_test)
+        y_pred_train = bagging_clf.predict(X_train)
+        accuracy_score_train = accuracy_score(y_true=y_train, y_pred=y_pred_train)
+        accuracy_score_test = accuracy_score(y_true=y_test, y_pred=y_pred_test)
+        lst_train_score.append(accuracy_score_train)
+        lst_test_score.append(accuracy_score_test)
+    plt.plot(estimator_arr, lst_train_score, color='b',label='Train Accuracy')
+    plt.plot(estimator_arr, lst_test_score, color='g',label='Test Accuracy')
+    plt.xlabel('Number of Estimators')
+    plt.ylabel('Accuracy Score')
+    plt.legend()
+    plt.show() 

q02_stacking_clf/build.py

@@ -15,4 +15,29 @@
 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=9)
 
 # Write your code here
+def stacking_clf(model, X_train, y_train, X_test, y_test):
+    X_train_meta = pd.DataFrame()
+    X_test_meta = pd.DataFrame()
+    for model_ in model:
+        # fit the models passed to method, using X_train and y_train
+        model_.fit(X_train,y_train)
+        # create train dataframe for Meta Classifier using models passed to the method
+        # predict the probabilties on train (mlxtend library does not use probabilities
+        # actual classes and hence the accuracy score using mlxtend is 0.74054054054054053)
+        # also we do not need to consider class 0 and class 1 probability in this case but
+        # test case is written such tht this implementation of the method will pass
+        df_meta_train = pd.DataFrame(model_.predict_proba(X_train))
+        X_train_meta = pd.concat([X_train_meta, df_meta_train],axis=1)
 
+        # create test dataframe for Meta Classifier using models passed to the method
+        # predict the probabilties on test
+        df_meta_test = pd.DataFrame(model_.predict_proba(X_test))
+        X_test_meta = pd.concat([X_test_meta, df_meta_test],axis=1)
+
+    # fit metaclassifier using Logistic
+    meta_logcf = LogisticRegression(random_state=9)
+    meta_logcf.fit(X_train_meta,y_train)
+    # Predict using metaclassifier using Logistic
+    y_pred_meta_test = meta_logcf.predict(X_test_meta)
+    acc_score = accuracy_score(y_true=y_test, y_pred=y_pred_meta_test)
+    return acc_score