Применение логистической регрессии в кредитном скоринге

sum()# Label encoding...data.replace({'Loan_Status':{'N':0,'Y': 1}},inplace = True)data.head()data['Dependents'].value_counts()data.replace({'Dependents': {'3+':4}},inplace = True)data['Dependents'].value_counts()data.head()sns.countplot(data= data,x = 'Education',hue = 'Loan_Status')sns.countplot(data= data,x = 'Married',hue = 'Loan_Status')sns.countplot(data= data,x = 'Dependents',hue = 'Loan_Status')sns.countplot(data= data,x = 'Self_Employed',hue = 'Loan_Status')sns.countplot(data = data,x = 'Gender',hue = 'Loan_Status')sns.countplot(data= data,x = 'Property_Area',hue = 'Loan_Status')sns.countplot(data= data,x = 'Credit_History',hue = 'Loan_Status')data.replace({'Married':{'No':0,'Yes': 1}},inplace = True)data.head()data.replace({'Education':{'Not Graduate':0,'Graduate': 1}},inplace = True)data.replace({'Self_Employed':{'No':0,'Yes': 1}},inplace = True)data.replace({'Gender':{'Female':0,'Male': 1}},inplace = True)data.replace({'Property_Area':{'Rural':0,'Urban': 1,'Semiurban':2}},inplace = True)data.head()plt.figure(figsize=(15,6))sns.heatmap(data.corr(),annot = True)sns.countplot(data= data,x = 'Loan_Amount_Term',hue = 'Loan_Status')X = data.drop(columns = ['Loan_Status','Loan_ID'],axis =1)y = data['Loan_Status']from sklearn.model_selection import train_test_splitX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)from sklearn.linear_model import LogisticRegressionlm = LogisticRegression()lm.fit(X_train,y_train)predictions = lm.predict(X_test)predictionsfrom sklearn.metrics import confusion_matrix,classification_report,accuracy_scoreprint(confusion_matrix(y_test,predictions))print(classification_report(y_test,predictions))accuracy_score(y_test,predictions)from sklearn.model_selection import train_test_splitX_train, X_test, Y_train, Y_test = train_test_split(X, y, test_size = 0.20, random_state = 0)from sklearn.preprocessing import StandardScalersc = StandardScaler()X_train = sc.fit_transform(X_train)X_test = sc.transform(X_test)from sklearn.preprocessing import StandardScalersc = StandardScaler()X_train = sc.fit_transform(X_train)X_test = sc.transform(X_test)#Using Logistic Regression Algorithm to the Training Setfrom sklearn.linear_model import LogisticRegressionlog = LogisticRegression(random_state = 0)print(log.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', log.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, log.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, log.predict(X_test)))print(F'Precision:', precision_score(Y_test, log.predict(X_test)))print(F'Recall:', recall_score(Y_test, log.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, log.predict(X_test)))#Check Roc Auc Scoreprint( F'RocAuc Score:',roc_auc_score(Y_test, log.predict(X_test)) )print( F'Balanced Accuracy Score:',balanced_accuracy_score(Y_test, log.predict(X_test)) ) print( F'Confusion Matrix:',confusion_matrix(Y_test, log.predict(X_test)) )print()#Print a new line# ROC CURVE#plot_roc_curve(log, X_test, Y_test)#plt.title('ROC Curve')#plt.plot([0, 1], [0, 1], 'r--')#plt.show()#Using KNeighborsClassifier Method of neighbors class to use Nearest Neighbor algorithmfrom sklearn.neighbors import KNeighborsClassifierknn = KNeighborsClassifier(n_neighbors = 5, metric = 'minkowski', p = 2)print(knn.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', knn.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, knn.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, knn.predict(X_test)))print(F'Precision:', precision_score(Y_test, knn.predict(X_test)))print(F'Recall:', recall_score(Y_test, knn.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, knn.predict(X_test)))#Check Roc Auc Scoreprint( F'RocAuc Score:',roc_auc_score(Y_test, knn.predict(X_test)) )print( F'Balanced Accuracy Score:',balanced_accuracy_score(Y_test, knn.predict(X_test)) ) print( F'Confusion Matrix:',confusion_matrix(Y_test, knn.predict(X_test)) )print()#Print a new line#Using GaussianNB method of naïve_bayes class to use Naïve Bayes Algorithmfrom sklearn.naive_bayes import GaussianNBgauss = GaussianNB()print(gauss.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', gauss.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, gauss.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, gauss.predict(X_test)))print(F'Precision:', precision_score(Y_test, gauss.predict(X_test)))print(F'Recall:', recall_score(Y_test, gauss.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, gauss.predict(X_test)))#Check Roc Auc Scoreprint( F'RocAuc Score:',roc_auc_score(Y_test, gauss.predict(X_test)) )print( F'Balanced Accuracy Score:',balanced_accuracy_score(Y_test, gauss.predict(X_test)) ) print( F'Confusion Matrix:',confusion_matrix(Y_test, gauss.predict(X_test)) )print()#Print a new line#Using DecisionTreeClassifier of tree class to use Decision Tree Algorithmfrom sklearn.tree import DecisionTreeClassifiertree =DecisionTreeClassifier(criterion = 'entropy', random_state = 0)print(tree.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', tree.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, tree.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, tree.predict(X_test)))print(F'Precision:', precision_score(Y_test, tree.predict(X_test)))print(F'Recall:', recall_score(Y_test, tree.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, tree.predict(X_test)))#Check Roc Auc Scoreprint( F'RocAuc Score:',roc_auc_score(Y_test, tree.predict(X_test)) )print( F'Balanced Accuracy Score:',balanced_accuracy_score(Y_test, tree.predict(X_test)) ) print( F'Confusion Matrix:',confusion_matrix(Y_test, tree.predict(X_test)) )print()#Print a new line#Using RandomForestClassifier method of ensemble class to use Random Forest Classification algorithmfrom sklearn.ensemble import RandomForestClassifierforest = RandomForestClassifier(n_estimators = 10, criterion = 'entropy', random_state = 10)print(forest.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', forest.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, forest.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, forest.predict(X_test)))print(F'Precision:', precision_score(Y_test, forest.predict(X_test)))print(F'Recall:', recall_score(Y_test, forest.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, forest.predict(X_test)))#Using xgboostr method of ensemble class to use xgboostClassification algorithmimport xgboost as xgbx_gb= xgb.XGBClassifier(objective="binary:logistic", n_estimators=150, random_state=42, eval_metric=["auc", "error", "error@0.6"])print(x_gb.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', x_gb.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, x_gb.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, x_gb.predict(X_test)))print(F'Precision:', precision_score(Y_test, x_gb.predict(X_test)))print(F'Recall:', recall_score(Y_test, x_gb.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, x_gb.predict(X_test)))#Check Roc Auc Scoreprint( F'RocAuc Score:',roc_auc_score(Y_test, x_gb.predict(X_test)) )print( F'Balanced Accuracy Score:',balanced_accuracy_score(Y_test, x_gb.predict(X_test)) ) print( F'Confusion Matrix:',confusion_matrix(Y_test, x_gb.predict(X_test)) )print()#Print a new line!pip install catboost#Using CatBoostClassifier method of ensemble class to use CatBoostClassification algorithmfrom catboost import CatBoostClassifierCatBoost = CatBoostClassifier(iterations=500,random_state=42,learning_rate=0.1)print(CatBoost.fit(X_train, Y_train))print('[0]Logistic Regression Training Accuracy:', CatBoost.score(X_train, Y_train))#Check Accuracy precision, recall, f1-scoreprint( classification_report(Y_test, CatBoost.predict(X_test)) )#Another way to get the models accuracy on the test dataprint(F'Accuracy:',accuracy_score(Y_test, CatBoost.predict(X_test)))print(F'Precision:', precision_score(Y_test, CatBoost.predict(X_test)))print(F'Recall:', recall_score(Y_test, CatBoost.predict(X_test)))print(F'F1 Score:', f1_score(Y_test, CatBoost.predict(X_test)))#Check Roc Auc Scoreprint( F'RocAuc Score:',roc_auc_score(Y_test, CatBoost.predict(X_test)) )print( F'Balanced Accuracy Score:',balanced_accuracy_score(Y_test, CatBoost.predict(X_test)) ) print( F'Confusion Matrix:',confusion_matrix(Y_test, CatBoost.predict(X_test)) )print()#Print a new line#Accuracy Scoreacc_1 = 0.87 acc_2 = 0.93acc_3 = 0.86acc_4 = 0.95acc_5 = 0.96 acc_6 = 0.96acc_7 = 0.97results = pd.DataFrame([["Logistic Regression",acc_1],["K-Nearest Neighbor",acc_2],["GaussianNB",acc_3], ["DecisionTreeClassifier",acc_4],["RandomForestClassifie",acc_5], ["xgboost",acc_6],["CatBoostClassifiert",acc_7]], columns = ["Models","Accuracy Score"]).sort_values(by='Accuracy Score',ascending=False)results.style.background_gradient(cmap='Blues')data.columnsX = data.drop(columns = ['Loan_ID','Loan_Amount_Term', 'Loan_Status'],axis =1)y = data['Loan_Status']X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)lm = LogisticRegression()lm.fit(X_train,y_train)predictions = lm.predict(X_test)accuracy_score(y_test,predictions)X = data.drop(columns = ['Loan_ID','Loan_Amount_Term', 'Loan_Status','Self_Employed'],axis =1)y = data['Loan_Status']X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)lm = LogisticRegression()lm.fit(X_train,y_train)predictions = lm.predict(X_test)accuracy_score(y_test,predictions)X = data.drop(columns = ['Loan_ID','Loan_Amount_Term', 'Loan_Status','LoanAmount','Self_Employed', 'ApplicantIncome', 'CoapplicantIncome'],axis =1)y = data['Loan_Status']X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)lm = LogisticRegression()lm.fit(X_train,y_train)predictions = lm.predict(X_test)accuracy_score(y_test,predictions)X = data.drop(columns = ['Loan_ID','Loan_Amount_Term', 'Loan_Status','LoanAmount','Self_Employed', 'ApplicantIncome', 'CoapplicantIncome', 'Gender'],axis =1)y = data['Loan_Status']X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)lm = LogisticRegression()lm.fit(X_train,y_train)predictions = lm.predict(X_test)accuracy_score(y_test,predictions)