README.md

---
language:
- en
size_categories:
- 100K<n<1M
task_categories:
- tabular-classification
task_ids:
- multi-class-classification
tags:
- credit-score
- finance
- banking
- tabular
- classification
pretty_name: Credit Score Classification Dataset
---

# 💳 Credit Score Classification Dataset

A comprehensive dataset for predicting customer credit scores into three categories: **Good**, **Standard**, and **Poor**.

## Dataset Description

This dataset contains customer financial information and behavioral patterns used for credit score classification. It includes various features related to credit history, payment behavior, and financial metrics.

### Dataset Summary

| Property | Value |
|----------|-------|
| **Total Samples** | ~100,000+ |
| **Features** | 22 (17 numerical + 5 categorical) |
| **Target Classes** | Good, Standard, Poor |
| **Format** | CSV |
| **Language** | English |

## Dataset Structure

### Data Files

| File | Description | Size |
|------|-------------|------|
| `train.csv` | Training dataset | ~31 MB |
| `test.csv` | Test dataset | ~15 MB |

### Features

#### Numerical Features (17)

| Feature | Description | Data Type |
|---------|-------------|-----------|
| `Age` | Customer's age in years | Integer |
| `Annual_Income` | Yearly income | Float |
| `Monthly_Inhand_Salary` | Monthly take-home salary | Float |
| `Num_Bank_Accounts` | Number of bank accounts owned | Integer |
| `Num_Credit_Card` | Number of credit cards | Integer |
| `Interest_Rate` | Average interest rate on credit | Integer |
| `Num_of_Loan` | Number of active loans | Integer |
| `Delay_from_due_date` | Average payment delay in days | Integer |
| `Num_of_Delayed_Payment` | Count of delayed payments | Integer |
| `Changed_Credit_Limit` | Credit limit change percentage | Float |
| `Num_Credit_Inquiries` | Number of credit inquiries | Integer |
| `Outstanding_Debt` | Total outstanding debt amount | Float |
| `Credit_Utilization_Ratio` | Credit utilization percentage | Float |
| `Credit_History_Age_Months` | Length of credit history in months | Integer |
| `Total_EMI_per_month` | Monthly EMI payments | Float |
| `Amount_invested_monthly` | Monthly investment amount | Float |
| `Monthly_Balance` | Average monthly balance | Float |

#### Categorical Features (5)

| Feature | Description | Categories |
|---------|-------------|------------|
| `Month` | Month of record | January - December |
| `Occupation` | Employment type | Accountant, Architect, Developer, Doctor, Engineer, Entrepreneur, Journalist, Lawyer, Manager, Mechanic, Media_Manager, Musician, Scientist, Teacher, Writer |
| `Credit_Mix` | Types of credit accounts | Bad, Good, Standard |
| `Payment_of_Min_Amount` | Minimum payment behavior | Yes, No, NM |
| `Payment_Behaviour` | Spending patterns | High_spent_Large_value_payments, High_spent_Medium_value_payments, High_spent_Small_value_payments, Low_spent_Large_value_payments, Low_spent_Medium_value_payments, Low_spent_Small_value_payments |

#### Target Variable

| Feature | Description | Classes |
|---------|-------------|---------|
| `Credit_Score` | Credit score classification | Good, Standard, Poor |

## Dataset Statistics

### Class Distribution

| Class | Description |
|-------|-------------|
| **Good** | Customers with excellent credit profiles |
| **Standard** | Customers with average credit profiles |
| **Poor** | Customers with concerning credit profiles |

### Feature Statistics (Approximate)

| Feature | Min | Max | Mean |
|---------|-----|-----|------|
| Age | 14 | 100 | ~35 |
| Annual_Income | 0 | 500,000 | ~50,000 |
| Num_Bank_Accounts | 0 | 20 | ~5 |
| Credit_Utilization_Ratio | 0% | 100% | ~30% |
| Credit_History_Age_Months | 0 | 500 | ~200 |

## Usage

### Loading with Pandas

```python
import pandas as pd

# Load training data
train_df = pd.read_csv('train.csv')
print(f"Training samples: {len(train_df)}")
print(f"Features: {train_df.columns.tolist()}")

# Load test data
test_df = pd.read_csv('test.csv')
print(f"Test samples: {len(test_df)}")
```

### Basic Exploration

```python
# Check class distribution
print(train_df['Credit_Score'].value_counts())

# Check for missing values
print(train_df.isnull().sum())

# Statistical summary
print(train_df.describe())
```

## Data Preprocessing

The following preprocessing steps are recommended:

1. **Handle Missing Values**: Some columns may contain missing or placeholder values
2. **Clean Categorical Data**: Handle special characters in categorical columns
3. **Feature Scaling**: Apply StandardScaler to numerical features
4. **Encoding**: Use OneHotEncoder for categorical features, LabelEncoder for target

### Example Preprocessing

```python
from sklearn.preprocessing import StandardScaler, LabelEncoder, OneHotEncoder
import pandas as pd

# Numerical columns
numerical_cols = ['Age', 'Annual_Income', 'Monthly_Inhand_Salary', 
                  'Num_Bank_Accounts', 'Num_Credit_Card', 'Interest_Rate',
                  'Num_of_Loan', 'Delay_from_due_date', 'Num_of_Delayed_Payment',
                  'Changed_Credit_Limit', 'Num_Credit_Inquiries', 'Outstanding_Debt',
                  'Credit_Utilization_Ratio', 'Credit_History_Age_Months',
                  'Total_EMI_per_month', 'Amount_invested_monthly', 'Monthly_Balance']

# Categorical columns
categorical_cols = ['Month', 'Occupation', 'Credit_Mix', 
                    'Payment_of_Min_Amount', 'Payment_Behaviour']

# Scale numerical features
scaler = StandardScaler()
X_numerical = scaler.fit_transform(train_df[numerical_cols])

# Encode target
label_encoder = LabelEncoder()
y = label_encoder.fit_transform(train_df['Credit_Score'])
```

## Considerations for Using the Data

### Data Quality Issues

- Some columns may contain placeholder values (e.g., `_`, `________`, `!@9#%8`)
- Credit history age may need conversion from text format
- Some numerical columns may have outliers

### Ethical Considerations

⚠️ **Important**: When using this data for credit scoring models:

- Be aware of potential biases in the data
- Ensure compliance with local financial regulations
- Credit decisions should not be based solely on automated predictions
- Provide transparency and explanations for credit decisions

### Recommended Cleaning Steps

```python
# Example: Handle placeholder values
placeholders = ['_', '________', '!@9#%8', 'NM']
for col in categorical_cols:
    train_df[col] = train_df[col].replace(placeholders, 'Unknown')
```

## Related Models

- **Model**: [AdityaaXD/credit-score-classifier](https://huggingface.co/AdityaaXD/credit-score-classifier)
- **GitHub**: [Credit-Score-Classification](https://github.com/ADITYA-tp01/Credit-Score-Clasification)

## Citation

```bibtex
@dataset{credit-score-dataset,
  author = {Aditya},
  title = {Credit Score Classification Dataset},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/AdityaaXD/credit-score-dataset}
}
```

## Contact

- **Hugging Face**: [@AdityaaXD](https://huggingface.co/AdityaaXD)
- **GitHub**: [@ADITYA-tp01](https://github.com/ADITYA-tp01)