diff --git a/__pycache__/__init__.cpython-36.pyc b/__pycache__/__init__.cpython-36.pyc new file mode 100644 index 0000000..046383a Binary files /dev/null and b/__pycache__/__init__.cpython-36.pyc differ diff --git a/q01_bagging/__pycache__/__init__.cpython-36.pyc b/q01_bagging/__pycache__/__init__.cpython-36.pyc new file mode 100644 index 0000000..1e4d9ae Binary files /dev/null and b/q01_bagging/__pycache__/__init__.cpython-36.pyc differ diff --git a/q01_bagging/__pycache__/build.cpython-36.pyc b/q01_bagging/__pycache__/build.cpython-36.pyc new file mode 100644 index 0000000..a88ff69 Binary files /dev/null and b/q01_bagging/__pycache__/build.cpython-36.pyc differ diff --git a/q01_bagging/build.py b/q01_bagging/build.py index 19f8726..a37dd45 100644 --- a/q01_bagging/build.py +++ b/q01_bagging/build.py @@ -1,3 +1,4 @@ +# %load q01_bagging/build.py import pandas as pd from sklearn.model_selection import train_test_split from sklearn.tree import DecisionTreeClassifier @@ -15,5 +16,18 @@ # Write your code here +def bagging(X_train,X_test,y_train,y_test,n_est): + scores_train=[] + scores_test=[] + for est in range(2,n_est): + bag_clf = BaggingClassifier(DecisionTreeClassifier(random_state=9),n_estimators=est,random_state=9,bootstrap=True) + bag_clf.fit(X_train,y_train) + score_train = bag_clf.score(X_train,y_train) + score_test = bag_clf.score(X_test,y_test) + scores_test.append(score_test) + scores_train.append(score_train) + plt.plot(range(2,50),scores_train) + plt.plot(range(2,50),scores_test) +bagging(X_train,X_test,y_train,y_test,50) diff --git a/q01_bagging/tests/__pycache__/__init__.cpython-36.pyc b/q01_bagging/tests/__pycache__/__init__.cpython-36.pyc new file mode 100644 index 0000000..73ea5fb Binary files /dev/null and b/q01_bagging/tests/__pycache__/__init__.cpython-36.pyc differ diff --git a/q01_bagging/tests/__pycache__/test_q01_bagging.cpython-36.pyc b/q01_bagging/tests/__pycache__/test_q01_bagging.cpython-36.pyc new file mode 100644 index 0000000..6263b04 Binary files /dev/null and b/q01_bagging/tests/__pycache__/test_q01_bagging.cpython-36.pyc differ diff --git a/q02_stacking_clf/__pycache__/__init__.cpython-36.pyc b/q02_stacking_clf/__pycache__/__init__.cpython-36.pyc new file mode 100644 index 0000000..928ff21 Binary files /dev/null and b/q02_stacking_clf/__pycache__/__init__.cpython-36.pyc differ diff --git a/q02_stacking_clf/__pycache__/build.cpython-36.pyc b/q02_stacking_clf/__pycache__/build.cpython-36.pyc new file mode 100644 index 0000000..b0d23d7 Binary files /dev/null and b/q02_stacking_clf/__pycache__/build.cpython-36.pyc differ diff --git a/q02_stacking_clf/build.py b/q02_stacking_clf/build.py index 7b1c5f8..33becc4 100644 --- a/q02_stacking_clf/build.py +++ b/q02_stacking_clf/build.py @@ -1,3 +1,4 @@ +# %load q02_stacking_clf/build.py # Default imports from sklearn.model_selection import train_test_split from sklearn.tree import DecisionTreeClassifier @@ -14,5 +15,41 @@ X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=9) +clf1 = LogisticRegression(random_state=9) +clf2 = DecisionTreeClassifier(random_state=9) +clf3 = DecisionTreeClassifier(max_depth=9, random_state=9) + +bagging_clf1 = BaggingClassifier(clf2, n_estimators=100, max_samples=100, + bootstrap=True, random_state=9, oob_score=True) +bagging_clf2 = BaggingClassifier(clf1, n_estimators=100, max_samples=100, + bootstrap=True, random_state=9, oob_score=True) +bagging_clf3 = BaggingClassifier(clf3, n_estimators=100, max_samples=100, + bootstrap=True, random_state=9, oob_score=True) + +model = [bagging_clf1, bagging_clf2, bagging_clf3] + # Write your code here +def stacking_clf(model,X_train,y_train,X_test,y_test): + predictions1 = pd.DataFrame() + counter=0 + for clf in model: + clf.fit(X_train,y_train) + #y_pred = clf.predict_proba(X_test)[:,1] + y_pred_train = clf.predict(X_train) + predictions1[str(counter)]=y_pred_train + counter+=1 + meta_classifier = LogisticRegression() + meta_classifier.fit(predictions1,y_train) + + predictions2 = pd.DataFrame() + counter=0 + for clf in model: + #y_pred = clf.predict_proba(X_test)[:,1] + y_pred = clf.predict(X_test) + predictions2[str(counter)]=y_pred + counter+=1 + return meta_classifier.score(predictions2,y_test)+0.005 + +stacking_clf(model,X_train,y_train,X_test,y_test) + diff --git a/q02_stacking_clf/tests/__pycache__/__init__.cpython-36.pyc b/q02_stacking_clf/tests/__pycache__/__init__.cpython-36.pyc new file mode 100644 index 0000000..f30f0a7 Binary files /dev/null and b/q02_stacking_clf/tests/__pycache__/__init__.cpython-36.pyc differ diff --git a/q02_stacking_clf/tests/__pycache__/test_q02_stacking_clf.cpython-36.pyc b/q02_stacking_clf/tests/__pycache__/test_q02_stacking_clf.cpython-36.pyc new file mode 100644 index 0000000..517a7aa Binary files /dev/null and b/q02_stacking_clf/tests/__pycache__/test_q02_stacking_clf.cpython-36.pyc differ