Credit Risk Classification

Overview

This project focuses on building a binary classification model to assess the creditworthiness of borrowers. Using logistic regression, we predicted loan status (0 = healthy, 1 = high-risk) across 20,000 loan applications.

---

Dataset Description

File Path	Rows	Columns	Notes
Resources/lending_data.csv	20,000	8	7 feature columns + 1 target label (loan_status)

Features

Column	Type	Description
loan_size	float	Principal loan amount
interest_rate	float	Annual interest rate (%)
borrower_income	float	Borrower's yearly income (USD)
debt_to_income_ratio	float	Total debt divided by income
open_accounts	int	Number of open credit lines
derogatory_marks	int	Negative credit events
total_debt	float	Total current outstanding debt
loan_status	0 / 1	Target label: 0 = healthy, 1 = high-risk

---

Methodology

1. Separating Features and Labels

X = df.drop(columns=["loan_status"])
y = df["loan_status"]

2. Splitting into Training & Test Sets

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.25, random_state=1
)

3. Training the Logistic Regression Model

from sklearn.linear_model import LogisticRegression

model = LogisticRegression(max_iter=200, random_state=1)
model.fit(X_train, y_train)

4. Making Predictions and Evaluating Model Performance

from sklearn.metrics import confusion_matrix, classification_report

y_pred = model.predict(X_test)
cm = confusion_matrix(y_test, y_pred)
cr = classification_report(y_test, y_pred)

print("Confusion Matrix:")
print(cm)

print("Classification Report:")
print(cr)

---

Model Evaluation

Confusion Matrix

[[18558   207]
 [   37   582]]

• 18,558 True Negatives (correctly predicted healthy loans)

• 207 False Positives (predicted risky loan but was healthy)

• 37 False Negatives (predicted healthy loan but was risky)

• 582 True Positives (correctly predicted risky loans)

  The model rarely missed risky loans (only 37 false negatives), which is crucial in financial decision-making.

---

Classification Report

              precision    recall  f1-score   support

           0       1.00      0.99      1.00     18765
           1       0.84      0.94      0.89       619

    accuracy                           0.99     19384
   macro avg       0.92      0.97      0.94     19384
weighted avg       0.99      0.99      0.99     19384

• Precision (1.0): The model made very few false predictions overall.
• Recall (94%) for risky loans: High sensitivity to identifying default-prone applicants.
• F1-score (0.89): Balanced performance between precision and recall for high-risk loans.
• Accuracy (99%): Strong classification across nearly 20,000 samples.

---

Key Takeaways

• The logistic regression model demonstrated exceptional accuracy (99%).
• It correctly identified most risky loans with high recall and very few false negatives (only 37).
• The model’s performance makes it a viable option for early-stage loan screening and risk assessment.

---

Recommendations for Future Improvements

• Threshold Tuning: Adjust decision boundary to further optimize recall or precision based on business needs.
• Model Comparison: Try ensemble classifiers (Random Forest, XGBoost) for potential performance gains.
• Cross-Validation: Validate model stability across different subsets of the data.
• Feature Importance Analysis: Gain interpretability into what features influence predictions most.

---