Credit Risk Prediction Using Decision Trees (3MTT Assignment Task 2)

This project focuses on developing a predictive model for classifying credit risk using a dataset of 100,000 customer records and 24 features. The goal is to predict whether a customer has a Poor or Standard credit score using Decision Tree classifiers.

Step 1: Import and Explore Dataset

The dataset includes features like Age, Occupation, Annual Income, Monthly Balance, Credit Mix, and Payment Behaviour. After exploring, there were no missing values.

The dataset is fairly balanced: 53% Standard and 47% Poor. A balanced dataset is crucial as it ensures that the model does not favor one class over the other and produces reliable predictions for both “Poor” and “Standard” credit scores.

Step 2: Transform Categorical Features

Categorical attributes such as Month, Occupation, Credit Mix, Payment of Minimum Amount, and Payment Behaviour were numerically encoded for the Decision Tree.

Feature Original Value Encoded Value
Credit_Mix Good 1
Credit_Mix Standard 2
Credit_Mix Poor 3

Step 3: Original Decision Tree Model

The Decision Tree classifier was trained on 22 numeric features. Performance on the test set:

  • Accuracy: 0.77
  • F1-score: 0.75–0.78
  • ROC AUC: 0.7686

Confusion Matrix:

True \ Predicted Poor Standard
Poor 7053 2349
Standard 2258 8340

Step 4: Hyperparameter Tuning

Parameters tuned:

  • max_depth: [3, 5, 7]
  • min_samples_leaf: [5, 10]

Best parameters found: max_depth = 7, min_samples_leaf = 10. This ensures the tree captures patterns without overfitting.

Step 5: Tuned Decision Tree Model

Performance on test set after tuning:

  • Accuracy: 0.74
  • F1-score: 0.74 (both classes)
  • ROC AUC: 0.8109

Confusion Matrix:

True \ Predicted Poor Standard
Poor 7370 2032
Standard 3111 7487

Step 6: Comparison of Original vs Tuned Models

Metric Original Tuned Class-Tuned Example
Accuracy 0.77 0.74 0.75
F1-score 0.75–0.78 0.74 0.75
ROC AUC 0.7686 0.8109 0.8226

Observations:

  • Hyperparameter tuning improved ROC AUC, indicating better class discrimination.
  • Precision and recall trade-offs shifted: tuned model favors the positive class slightly.
  • Maintaining a balanced dataset helped the model perform reliably across both Poor and Standard classes.
  • Overall, the tuned model generalizes better while controlling overfitting.

Step 7: Conclusion

Hyperparameter tuning in the Decision Tree model improved its ability to distinguish between Poor and Standard credit scores, despite a minor drop in overall accuracy. Using a balanced dataset, careful feature encoding, and confusion matrix analysis ensures that the predictive model can be trusted for real-world credit risk assessment.

Comments

Post a Comment

Popular posts from this blog

NR-V2X Mode 2 in ns-3 In 5G NR-V2X, vehicles communicate directly using sidelink. Two modes exist: Mode 1: gNB schedules sidelink resources. Mode 2: UEs autonomously select sidelink resources (no gNB needed). Mode 2 is especially important for vehicular networks (IoV), where cars must exchange safety messages without relying on infrastructure. In ns-3, the NrHelper class is used to: Create UE devices with sidelink PHY/MAC layers. Attach them to Bandwidth Parts (BWPs) and spectrum channels. Configure sidelink attributes such as: SidelinkMode = 2 (autonomous) SidelinkPeriod (resource pool periodicity) SidelinkSubchannelSize (RB grouping)
Read more

From DSRC to 5G NR-V2X: The Road Ahead for Connected Vehicles

Modern vehicles are no longer just machines on wheels. Through Vehicle-to-Everything (V2X) communication , they are becoming active participants in intelligent transport systems, enabling safer driving, smoother traffic, and progress toward automation. DSRC: The First Step The earliest V2X technology was Dedicated Short-Range Communications (DSRC) , based on IEEE 802.11p. Operating in the 5.9 GHz band , it allowed onboard units (OBUs) in vehicles to exchange safety and infotainment data with roadside units (RSUs) . While DSRC was effective for short-range messaging, it faced limits in coverage, data rate, and quality of service (QoS) . These challenges motivated researchers to explore new solutions. D2D and the Cellular Shift A key breakthrough came with Device-to-Device (D2D) communication , where vehicles exchange information directly without always relying on infrastructure. This reduces latency, eases network congestion, and helps maintain reliability in crow...
Read more

CTE 311: ENGINEER IN SOCIETY: CURRICULUM (20/21 SESSION)

GENERAL OBJECTIVES: On completion of this module, the student should be able to: 1. Understand the historical development of engineering and technology. 2. Know the cadres and pyramidal structure of the engineering family. 3. Understand roles of engineering personnel in the society. 4. Know the engineering professional associations in Nigeria. 5. Understand importance and process of engineering professional registration in Nigeria. 1.1 Identify, describe, technology and engineering in ancient Egypt, Rome, China, Europe, America 1.2 State developments from military to Civil Engineering . 1.3 State the industrial Revolution in England and its spread to Europe, Asia and Africa. 1.4 State Mo dern technology and eng ineering. 2.1 Identify the various cadres o f the engineering family . 2.2 State the ideal ratio of the different cadres of engineering personnel required for an engineering project team. 2.3 Identify the various engineering discipline. 2.4 Identify the pyramidal structure ...
Read more