Credit Limit Prediction using Regression

Predicting credit limits is crucial for financial institutions to assess risk and make informed lending decisions. This project focuses on using regression models to predict credit limits based on various customer attributes.

Preprocessing Data

Before training our model, we preprocess the data to ensure its quality and suitability for analysis. This includes removing irrelevant columns, handling duplicate data, encoding categorical features, missed values imputation, outlier detection, scaling, and splitting the data into training and testing sets.

Handling Missing Data

Missing data is a common issue in datasets and can significantly impact model performance. We employ strategies such as imputation based on mean or median values and leveraging relationships between features to fill missing data appropriately.

Handling Outliers

Outliers can skew model predictions and affect the overall accuracy. We identify and remove outliers using techniques like Local Outlier Factor (LOF) to ensure robust model training.

Feature Selection

Selecting the most relevant features is crucial for model efficiency and interpretability. We use techniques like correlation analysis and feature importance scores to select the most informative features for our regression model.

	Feature	Importance	VIF
0	Avg_Utilization_Ratio	0.420781	6.062172
1	Income_Category	0.239581	8.630447
2	Total_Revolving_Bal	0.163871	7.174995
3	Card_Category	0.061275	1.296993
4	Total_Trans_Amt	0.016861	8.821232
5	Total_Amt_Chng_Q4_Q1	0.015222	16.236073
6	Total_Ct_Chng_Q4_Q1	0.014307	15.424124
7	Total_Trans_Ct	0.012174	25.031854
8	Customer_Age	0.010757	76.824692
9	Months_on_book	0.010366	57.213571
10	Total_Relationship_Count	0.007021	7.831607
11	Education_Level	0.006195	3.251686
12	Contacts_Count_12_mon	0.005856	5.587047
13	Dependent_count	0.004754	4.209314
14	Months_Inactive_12_mon	0.004551	6.314962
15	Marital_Status	0.004440	8.258021
16	Gender	0.001987	4.977622

Model Training

We train regression models, such as RandomForestRegressor, on the preprocessed data to predict credit limits. We fine-tune model parameters and evaluate performance to ensure optimal results.

Various regression models are trained and evaluated, including:

• Random Forest Regression
• Linear Regression
• Ridge Regression
• Polynomial Regression

Evaluation

We evaluate model performance using metrics such as Mean Squared Error (MSE) and R-squared.

• Mean Squared Error (MSE): Measures the average squared difference between the actual and predicted values.
• R-squared (R2) Score: Represents the proportion of the variance in the dependent variable that is predictable from the independent variables.

Model	Train MSE	Train R2	Test MSE	Test R2
Random Forest	1.253357e+06	0.982717	1.266337e+07	0.851246
Polynomial Regression	1.012446e+07	0.860389	3.010598e+07	0.646350
Linear Regression	2.948030e+07	0.593483	3.622185e+07	0.574508
Ridge Regression	2.948209e+07	0.593458	3.627357e+07	0.573901

Conclusion

In conclusion, this project demonstrates the application of regression models for credit limit prediction. By preprocessing data, handling missing values and outliers, and selecting informative features, we build robust models that can assist financial institutions in making informed lending decisions.

Contributing

Contributions to this project are welcome! If you have any suggestions, improvements, or bug fixes, feel free to open an issue or submit a pull request.