README.md

---
license: mit
---

# ML Assignment 3 – Maha Qaiser
Dataset: California Housing from sklearn.datasets

## File Description
- `best_model.joblib`: Mini-Batch Linear Regression model
- `scaler.joblib`: StandardScaler object to preprocess input features
- `inference.py`: Script to load the model + scaler and run predictions with user input

## Model Overview
- **Model Type:** Mini-Batch Linear Regression
- **Features Used:**
  - Avg. Rooms
  - Avg. Bedrooms
  - Population
  - Household
  - Median Income
  - Latitude
  - Longitude
  - Housing Median Age
- **Regularization:** L2 (Ridge)
- **Early Stopping:** Applied during training

## How to Run Inference
### 1. Clone the repository or download the files:
```bash
git clone https://huggingface.co/mahaqj/ml_assignment_3
cd ml_assignment_3
```
### 2. Install dependencies:
```bash
pip install joblib numpy scikit-learn huggingface_hub
```
### 3. Run the script:
```bash
python inference.py
```

You’ll be prompted to enter the following features:
- Avg. Rooms
- Avg. Bedrooms
- Population
- Household
- Median Income
- Latitude
- Longitude
- Housing Median Age

The model will return the predicted housing value.

## Loading the Model in Python
```python
import joblib
import requests
from io import BytesIO

# urls to both model and scaler
model_url = "https://huggingface.co/mahaqj/ml_assignment_3/resolve/main/best_model.joblib"
scaler_url = "https://huggingface.co/mahaqj/ml_assignment_3/resolve/main/scaler.joblib"

# download and load model
model_bytes = BytesIO(requests.get(model_url).content)
model = joblib.load(model_bytes)

# download and load scaler
scaler_bytes = BytesIO(requests.get(scaler_url).content)
scaler = joblib.load(scaler_bytes)
```