apps/fastapi_variants/README.md

# FastAPI Multi-Variant API for Milk Spoilage Classification

This API supports 10 model variants optimized for different data availability scenarios.

## Features

- **10 Model Variants** with test accuracies ranging from 62.8% to 95.8%
- **Automatic Feature Validation** - API validates you provide required features for selected variant
- **RESTful API** with comprehensive OpenAPI documentation
- **Custom GPT Ready** - Designed for seamless ChatGPT integration

## Quick Start

### Local Development

```bash
# Install dependencies
pip install -r requirements.txt

# Run the server
python app.py
```

The API will be available at `http://localhost:7860`

### Docker Deployment

```bash
docker build -t milk-spoilage-api .
docker run -p 7860:7860 milk-spoilage-api
```

## API Endpoints

### GET /variants
List all available model variants with metadata

### POST /predict
Make a prediction using the specified model variant

**Request Body:**
```json
{
  "spc_d7": 2.1,
  "spc_d14": 4.7,
  "spc_d21": 6.4,
  "tgn_d7": 1.0,
  "tgn_d14": 3.7,
  "tgn_d21": 5.3,
  "model_variant": "baseline"
}
```

**Response:**
```json
{
  "prediction": "PPC",
  "probabilities": {
    "PPC": 0.97,
    "no spoilage": 0.02,
    "spore spoilage": 0.01
  },
  "confidence": 0.97,
  "variant_used": {
    "variant_id": "baseline",
    "name": "Baseline (All Features)",
    "description": "Uses all 6 microbiological measurements",
    "features": ["SPC_D7", "SPC_D14", "SPC_D21", "TGN_D7", "TGN_D14", "TGN_D21"],
    "test_accuracy": 0.9576
  }
}
```

## Model Variants

| Variant | Test Acc | Features Required |
|---------|----------|-------------------|
| baseline | 95.8% | All 6 features |
| scenario_1_days14_21 | 94.2% | SPC_D14, SPC_D21, TGN_D14, TGN_D21 |
| scenario_3_day21 | 93.7% | SPC_D21, TGN_D21 |
| scenario_4_day14 | 87.4% | SPC_D14, TGN_D14 |
| scenario_2_days7_14 | 87.3% | SPC_D7, SPC_D14, TGN_D7, TGN_D14 |
| scenario_6_spc_all | 78.3% | SPC_D7, SPC_D14, SPC_D21 |
| scenario_8_spc_7_14 | 73.3% | SPC_D7, SPC_D14 |
| scenario_9_tgn_7_14 | 73.1% | TGN_D7, TGN_D14 |
| scenario_7_tgn_all | 69.9% | TGN_D7, TGN_D14, TGN_D21 |
| scenario_5_day7 | 62.8% | SPC_D7, TGN_D7 |

## Custom GPT Integration

See `../../docs/CUSTOM_GPT_SETUP_MULTIVARIANT.md` for complete setup instructions.

## Example Usage

### Python

```python
import requests

url = "https://chenhaoq87-milkspoilageclassifier-api-variants.hf.space/predict"

# Using baseline model with all features
response = requests.post(url, json={
    "spc_d7": 2.1,
    "spc_d14": 4.7,
    "spc_d21": 6.4,
    "tgn_d7": 1.0,
    "tgn_d14": 3.7,
    "tgn_d21": 5.3,
    "model_variant": "baseline"
})

print(response.json())

# Using Day 21 only model (when you don't have earlier measurements)
response = requests.post(url, json={
    "spc_d21": 6.4,
    "tgn_d21": 5.3,
    "model_variant": "scenario_3_day21"
})

print(response.json())
```

### cURL

```bash
# List available variants
curl https://chenhaoq87-milkspoilageclassifier-api-variants.hf.space/variants

# Make prediction with baseline model
curl -X POST https://chenhaoq87-milkspoilageclassifier-api-variants.hf.space/predict \
  -H "Content-Type: application/json" \
  -d '{
    "spc_d7": 2.1,
    "spc_d14": 4.7,
    "spc_d21": 6.4,
    "tgn_d7": 1.0,
    "tgn_d14": 3.7,
    "tgn_d21": 5.3,
    "model_variant": "baseline"
  }'
```

## License

MIT