# Import pandas
import pandas as pd

# Read the CSV file
df = pd.read_csv("data_credit_risk.csv")
# df = pd.read_csv("https://raw.githubusercontent.com/ash322ash422/data/refs/heads/main/data_credit_risk.csv")


# Display the first 5 rows
df.head()

# Check the shape of the dataset

print("Shape of dataset:", df.shape)

# Display column names

print(df.columns.tolist())

df.info()

# Summary statistics for numerical columns

df.describe()

# Summary for categorical columns

df.describe(include='object')

# Check for missing values

print("missing value:\n", df.isnull().sum())

# VISUAL: MISSING VALUES

missing = df.isnull().sum()
missing = missing[missing > 0]

missing.plot(kind='bar')

plt.title('Missing Values by Column')
plt.ylabel('Count')
plt.show()

# Check percentage of missing values

missing_percent = df.isnull().mean() * 100
print(missing_percent)

# Fill missing values with median


df['person_emp_length'] = df['person_emp_length'].fillna(df['person_emp_length'].median())
df['loan_int_rate'] = df['loan_int_rate'].fillna(df['loan_int_rate'].median())

# Verify that missing values are gone
print(df.isnull().sum())

df['loan_status'].value_counts()

print("percentage:\n", df['loan_status'].value_counts(normalize=True)*100)

# VISUAL 1: TARGET VARIABLE DISTRIBUTION

import matplotlib.pyplot as plt

df['loan_status'].value_counts().sort_index().plot(
    kind='bar'
)

plt.xticks([0, 1], ['Safe (0)', 'Default (1)'], rotation=0)
plt.title('Loan Status Distribution')
plt.ylabel('Number of Customers')
plt.show()

default_by_grade = df.groupby('loan_grade')['loan_status'].mean()

default_by_grade.plot(kind='bar')

plt.title('Default Rate by Loan Grade')
plt.ylabel('Default Rate')
plt.xlabel('Loan Grade')
plt.ylim(0, 1)
plt.show()

plt.figure(figsize=(8, 4))

plt.hist(
    df[df['loan_status'] == 0]['person_income'],
    bins=30,
    alpha=0.6,
    label='Safe'
)

plt.hist(
    df[df['loan_status'] == 1]['person_income'],
    bins=30,
    alpha=0.6,
    label='Default'
)

plt.legend()
plt.title('Income Distribution by Loan Status')
plt.xlabel('Income')
plt.ylabel('Frequency')
plt.show()

column_outliers = ['person_income', 'loan_amnt', 'loan_int_rate'] 

for col in column_outliers:
    plt.figure(figsize=(6, 2))
    plt.boxplot(df[col].dropna(), vert=False)
    plt.title(col)
    plt.show()

for col in column_outliers:
    lower = df[col].quantile(0.01)
    upper = df[col].quantile(0.99)
    df[col] = df[col].clip(lower, upper)

for col in column_outliers:
    plt.figure(figsize=(6, 2))
    plt.boxplot(df[col].dropna(), vert=False)
    plt.title(col)
    plt.show()

# Target variable
y = df['loan_status']

# Input features (all columns except target)
X = df.drop('loan_status', axis=1)

print("Feature matrix shape:", X.shape)
print("Target vector shape:", y.shape)

# View first few rows
X.head()

# Numerical columns
num_cols = X.select_dtypes(include=['int64', 'float64']).columns.tolist()

# Categorical columns
cat_cols = X.select_dtypes(include=['object']).columns.tolist()

print("Numerical Columns:")
print(num_cols)

print("\nCategorical Columns:")
print(cat_cols)

# Convert categorical variables to dummy variables
X_encoded = pd.get_dummies(X, columns=cat_cols, drop_first=True)

print("Shape after encoding:", X_encoded.shape)
X_encoded.head()

X_encoded.columns

# Check Final Dataset
print("Final feature matrix shape:", X_encoded.shape)
print("Target distribution:")
print(y.value_counts(normalize=True))

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(
    X_encoded,      # Feature matrix
    y,              # Target variable
    test_size=0.20, # 20% for testing
    random_state=42,
    stratify=y      # Preserve class proportions
)

print("Training set shape:", X_train.shape)
print("Testing set shape :", X_test.shape)
print("Training target distribution:")
print(y_train.value_counts(normalize=True))

from sklearn.naive_bayes import GaussianNB

# Create model
model = GaussianNB()

# Train model
model.fit(X_train, y_train)

print("Model training complete.")

# Predicted class labels (0 or 1)
y_pred = model.predict(X_test)

# Predicted probabilities
# [:, 1] gives the probability of class 1 (loan default)
y_prob = model.predict_proba(X_test)[:, 1]

print("First 10 Predictions:")
print(y_pred[:10])

print("\nFirst 10 Probabilities of Default:")
print(y_prob[:10])

# VISUAL: HISTOGRAM OF PREDICTED PROBABILITIES


plt.figure(figsize=(8, 4))
plt.hist(y_prob, bins=30)

plt.title('Predicted Probability of Default')
plt.xlabel('Probability')
plt.ylabel('Number of Customers')
plt.show()

from sklearn.metrics import (
    accuracy_score,
    confusion_matrix,
    classification_report,
    roc_auc_score
)

# Accuracy
accuracy = accuracy_score(y_test, y_pred)
print("Accuracy:", round(accuracy, 4))

# Confusion Matrix
cm = confusion_matrix(y_test, y_pred)
print("\nConfusion Matrix:")
print(cm)

# Classification Report
print("\nClassification Report:")
print(classification_report(y_test, y_pred))

# ROC-AUC Score
auc = roc_auc_score(y_test, y_prob)
print("ROC-AUC Score:", round(auc, 4))

# VISUAL 6: ROC CURVE

from sklearn.metrics import roc_curve, roc_auc_score

fpr, tpr, thresholds = roc_curve(y_test, y_prob)
auc = roc_auc_score(y_test, y_prob)

plt.figure(figsize=(6, 5))
plt.plot(fpr, tpr, label=f'AUC = {auc:.3f}')
plt.plot([0, 1], [0, 1], '--')

plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('ROC Curve')
plt.legend()
plt.show()

# INTERPRET A FEW PREDICTIONS

# Create a comparison table
results = pd.DataFrame({
    "Actual": y_test.values,
    "Predicted": y_pred,
    "Probability_of_Default": y_prob
})

results.head(10)

results.sort_values(
    by="Probability_of_Default",
    ascending=False
).head(10)

# VISUAL : TOP N HIGHEST RISK CUSTOMERS
N=20

results = pd.DataFrame({
    'Actual': y_test.values,
    'Predicted': y_pred,
    'Probability_of_Default': y_prob
})

top_N = results.sort_values(
    by='Probability_of_Default',
    ascending=False
).head(N)

plt.figure(figsize=(8, 4))
plt.bar(
    range(len(top_N)),
    top_N['Probability_of_Default']
)

plt.xticks(
    range(len(top_N)),
    top_N.index,
    rotation=45
)

plt.ylim(0, 1)
plt.title(f'Top {N} Highest-Risk Customers')
plt.ylabel('Probability of Default')
plt.show()