# 
## File Structure

```
├── train_input.npy
├── test_input.npy
└── sample_submission.csv
```

## Detailed File Descriptions

### 1. **train_input.npy**
**Purpose**: Train the classification model 

**Structure**:
- **Format**: NumPy array (.npy)
- **Shape**: (N_train, 11) where N_train is the number of training pixels
- **Columns**:
  - **Columns 1-10**: Polarimetric scattering features (see **Feature Descriptions** below)
  - **Column 11**: Ground truth land cover label (integer: 1-15)


---

### 2. **test_input.npy**
**Purpose**: Test dataset for prediction

**Structure**:
- **Format**: NumPy array (.npy)
- **Shape**: (N_test, 10) where N_test is the number of test pixels
- **Columns**:
  - **Columns 1-10**: Polarimetric scattering features (identical feature set as training data)


---

### 3. **sample_submission.csv**
**Purpose**: Submission template

**Structure**:
- **Format**: CSV file
- **Columns**:
  - **Column 1 (ID)**: Pixel index (1 to N_test)
  - **Column 2 (Label)**: Predicted land cover class (integer: 1-15)

**Example**:
```csv
ID,Label
0,3
1,7
2,12
...
```


---

## 🔬 Feature Descriptions

The 10 polarimetric scattering features are extracted from quad-polarimetric SAR data:

| Feature Index | Feature Name | Description |
|--------------|--------------|-------------|
| 1 | C11 | Covariance matrix element (1,1) - HH intensity |
| 2 | C12 | Covariance matrix element (1,2) |
| 3 | C13 | Covariance matrix element (1,3) |
| 4 | C22 | Covariance matrix element (2,2) - HV intensity |
| 5 | C23 | Covariance matrix element (2,3) |
| 6 | C33 | Covariance matrix element (3,3) - VV intensity |
| 7 | SPAN | Total scattered power (C11+C22+C33) |
| 8 | Alpha (α) | Average scattering angle |
| 9 | Anisotropy (A) | Polarimetric anisotropy |
| 10 | Entropy (H) | Randomness of scattering mechanisms |


---