Credit Risk Estimation

Credit risk estimation is a critical task for financial
institutions to minimize losses from loan defaults. This paper
presents a comprehensive evaluation of multiple machine learning
models, including Logistic Regression, Random Forest, Naive
Bayes, Support Vector Machines (SVM), XGBoost, LightGBM,
and CatBoost, for predicting loan defaults. The models are
trained on a dataset containing demographic, financial, and loan-
related features. Preprocessing steps, including handling missing
values, one-hot encoding, and feature scaling, are discussed in
detail. The best-performing model, CatBoost, achieves a test
accuracy of 93.62% and an F1-score of 93.28%. We also dis-
cuss ethical considerations, such as fairness across demographic
subgroups, and provide insights into model interpretability and
deployment. The results demonstrate that ensemble methods, par-
ticularly CatBoost, are highly effective for credit risk estimation.

app.py

@@ -6,7 +6,7 @@ app = Flask(__name__)
 CORS(app)  # Enable CORS for all routes
 
 # Load your model
-model = pickle.load(open("models/best_model_catboost.pkl", "rb"))
+model = pickle.load(open("best_model_catboost.pkl", "rb"))
 
 @app.route('/predict', methods=['POST'])
 def predict():