AdityaaXD/credit-score-classifier

💳 Credit Score Classifier

A Random Forest Classifier trained to predict customer credit scores into three categories: Good, Standard, and Poor.

Model Description

This model analyzes customer financial data and behavioral patterns to classify their credit worthiness. It was trained on a comprehensive dataset containing financial metrics, payment history, and credit utilization patterns.

Model Details

Property	Value
Model Type	Random Forest Classifier
Framework	Scikit-learn
Number of Trees	100
Target Classes	Good, Standard, Poor
Input Features	17 numerical + 5 categorical

Intended Use

Primary Use Cases

• Credit Risk Assessment: Evaluate creditworthiness of loan applicants
• Financial Services: Automate preliminary credit screening
• Banking Applications: Support credit limit decisions

Out-of-Scope Use

• This model should not be the sole decision-maker for credit approvals
• Not intended for use without human oversight
• Should not be used for discriminatory purposes

How to Use

Installation

pip install huggingface_hub scikit-learn pandas numpy

Loading the Model

from huggingface_hub import hf_hub_download
import pickle

# Download model files
model_path = hf_hub_download(repo_id="AdityaaXD/credit-score-classifier", filename="models/final_model.pkl")
scaler_path = hf_hub_download(repo_id="AdityaaXD/credit-score-classifier", filename="models/scaler.pkl")
label_encoder_path = hf_hub_download(repo_id="AdityaaXD/credit-score-classifier", filename="models/label_encoder.pkl")
feature_info_path = hf_hub_download(repo_id="AdityaaXD/credit-score-classifier", filename="models/feature_info.pkl")

# Load the model
with open(model_path, "rb") as f:
    model = pickle.load(f)

with open(scaler_path, "rb") as f:
    scaler = pickle.load(f)

with open(label_encoder_path, "rb") as f:
    label_encoder = pickle.load(f)

Making Predictions

import pandas as pd
import numpy as np

# Example: Prepare your input data
numerical_features = {
    'Age': 30,
    'Annual_Income': 50000,
    'Monthly_Inhand_Salary': 4000,
    'Num_Bank_Accounts': 4,
    'Num_Credit_Card': 3,
    'Interest_Rate': 12,
    'Num_of_Loan': 2,
    'Delay_from_due_date': 5,
    'Num_of_Delayed_Payment': 3,
    'Changed_Credit_Limit': 8.0,
    'Num_Credit_Inquiries': 4,
    'Outstanding_Debt': 1200,
    'Credit_Utilization_Ratio': 28.5,
    'Credit_History_Age_Months': 180,
    'Total_EMI_per_month': 150,
    'Amount_invested_monthly': 200,
    'Monthly_Balance': 500
}

# Scale numerical features
num_df = pd.DataFrame([numerical_features])
scaled_features = scaler.transform(num_df)

# Make prediction (note: categorical features need one-hot encoding)
prediction = model.predict(scaled_features)
predicted_class = label_encoder.inverse_transform(prediction)
print(f"Predicted Credit Score: {predicted_class[0]}")

Training Data

The model was trained on a credit score dataset containing:

Feature Type	Count	Examples
Numerical	17	Age, Annual Income, Outstanding Debt, Credit Utilization
Categorical	5	Occupation, Credit Mix, Payment Behavior

Input Features

Numerical Features

• Age - Customer's age
• Annual_Income - Yearly income
• Monthly_Inhand_Salary - Monthly take-home salary
• Num_Bank_Accounts - Number of bank accounts
• Num_Credit_Card - Number of credit cards
• Interest_Rate - Average interest rate
• Num_of_Loan - Number of active loans
• Delay_from_due_date - Average payment delay (days)
• Num_of_Delayed_Payment - Count of delayed payments
• Changed_Credit_Limit - Credit limit changes
• Num_Credit_Inquiries - Number of credit inquiries
• Outstanding_Debt - Total outstanding debt
• Credit_Utilization_Ratio - Credit utilization percentage
• Credit_History_Age_Months - Credit history length
• Total_EMI_per_month - Monthly EMI payments
• Amount_invested_monthly - Monthly investments
• Monthly_Balance - Average monthly balance

Categorical Features

• Month - Month of record
• Occupation - Employment type
• Credit_Mix - Types of credit accounts
• Payment_of_Min_Amount - Minimum payment behavior
• Payment_Behaviour - Spending patterns

Model Files

File	Description
models/final_model.pkl	Trained Random Forest model
models/scaler.pkl	StandardScaler for numerical features
models/label_encoder.pkl	LabelEncoder for target classes
models/feature_info.pkl	Feature metadata and column names
models/onehot_encoder.pkl	OneHotEncoder for categorical features

Limitations

• Data Bias: Model performance depends on training data quality and may not generalize to all populations
• Feature Availability: Requires all 17 numerical and 5 categorical features for accurate predictions
• Temporal Drift: Financial patterns change over time; periodic retraining recommended
• Geographic Scope: Trained on specific regional data; may need adaptation for other regions

Ethical Considerations

⚠️ Important: This model is intended as a decision-support tool, not a replacement for human judgment.

• Always combine model predictions with human review
• Be aware of potential biases in credit scoring
• Ensure compliance with local financial regulations
• Provide explanations for credit decisions when required by law

Demo Application

Try the interactive Streamlit demo: Credit Score Classifier App

Citation

@misc{credit-score-classifier,
  author = {Aditya},
  title = {Credit Score Classification using Random Forest},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/AdityaaXD/credit-score-classifier}
}

Contact

• Hugging Face: @AdityaaXD
• GitHub: @ADITYA-tp01