Implemented SGD from scratch for 5 features like in Lesson 2. How do I optimize?

vishak · July 27, 2019, 6:03pm

As Jeremy does with a single feature in Lesson 2. I choose more features and tried to implement SGD from scratch but as the o/p post SGD compares with my initial parameters, there is still a lot of be desired.
I tweaked the number of loops and the LR but it didn’t change much.
Can someone please tell me what I’m missing here and why I can’t make my regression come up with better solutions?

github.com

VishakBharadwaj94/Machine_Learning/blob/master/my_sgd.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "my_sgd.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",

This file has been truncated. show original

Thank You!