Feature description

I noticed that the current linear_regression.py
Implementation is partially vectorized but still uses Python loops and manual parsing in some places. I think for a learning platform, we should include both naive and vectorized algorithms.