Linear Regression

This is Python implementation of the Linear Regression code written in Matlab/Octave in exercise files of Machine Learning course by Andrew Ng. It contains first three parts of the code till Gradient Descent algorithm implementation.

import getch
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

• The first library is getch which is used to implement press any key to continue in python. The function used is getch.getch()
• The second library used is our favorite pandas library for dataframes and importing big data
• Third one is matplotlib as you can see. It is used for visualization of the data by plotting it in form of graphs
• The last one is numpy for handling matrices and matrix opertaions

The first section in the machine-learning-ex1 was the warmupExercise where we were required to generate a 5X5 matrix with 1s in it diagonal

warmUpExercise

Python implementation

def warmUpExercise():
    A = np.eye(5)

    return A

Here numpy has in-built function eye() which returns the matrix according to the value of para,eter passed. The result is stored in variable and returned.

Matlab/Octave implementation

function A = warmUpExercise()
  A = [];
  A=eye(5);
end

Plotting Data

The second function in the file is plotData() it takes the data, theta and bool as input aand displays the graph.

Python implementation

def plotData(datum, theta=np.array([[0], [0]]), draw_line=False):
    X = datum.pop('X')  # taking the X column only
    Y = datum.pop('Y')  # taking the Y column only

    plt.scatter(X, Y, marker='x')  # Creating a scatter plot with X and Y

    if draw_line is True:  # Adding [1]s in X at position 0
        for elem in X:
            elem = [1, elem]
            newX.append(elem)

        X = newX
        X = np.array(X)
        plt.plot(X[:, 1], np.dot(X, theta), marker='_')

    plt.show()

    return X, Y

It returns the colums of the data provided by taking in the data.

• If we only have to plot data then we can simply use the first arguement and leave the rest blank. plotData(datum)
• If we want to draw the line after computing cost function then we can use the rest of the arguements, i.e., theta and the bool value e.g. plotData(data, np.array([[1],[2]]), True)
• The scatter() function plot scatter plot afer taking the required arguements.
• According to the bool value the if statement is executed and the line is drawn using X and the theta
• The plot is made and the columns are returned

Matlab/Octave implementation

function plotData(x, y)

figure;
  plot(x, y, "rx", "MarkerSize", 10);
  axis([4 24 -5 25]);
  xlabel("Population of City in 10,000s"); 
  ylabel("Profit in $10,000s");
end