Setup Guide

This guide will provide a step-by-step guide to make your machine ready to run Python with basic packages to get you started with machine learing.

Installing Python

Download PyCharm (community version). It is one of the widely used IDE for Python and would make coding easier.

Install HomeBrew a package manager for Mac OS by running the following commands in the terminal.

xcode-select --install

ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

Ensure Brew is correctly installed by running the following command.

brew doctor

Install Python 3.7 by running the following command.

brew install python3

After the installation check the version of the python and also its installation directory.

python3 --version
which python3

The ouput of the directory should be something like "/usr/local/bin/python3.7" or "/Library/Frameworks/Python.framework/Versions/3.7/bin/python3.7"

Getting Started

1. Start a new project in PyCharm. And push the dropdown of "Project Interpreter".
2. Choose the option of "New environment using Virtualenv" and check that the version of the "Base interpreter" is Python3.7. Check your project directory and click "Create". Virtualenv allows us to have a seprate environment for each project which allows for smooth development.
3. Create a new Python file either from the File menu or by right-clicking the project folder in the left widget.
4. Let's run a simple code to understand how to run and view output.

if __name__ == '__main__':
	a = "This is a string."
	b = 10
	c = 55
	result = b + c
	print ("Hello World")
	print ("The sum of two numbers: {}".format(result))

5. To run the above program either click the small green triangle besides the "if __name__ == '__main__':" or run the program from the Run menu.
6. Feel free to explore some of the basic Python tutorials by clicking here.

Solving our first ML Problem

We are gonna work on our first dataset by using simple linear regression. We would be using a fairly small dataset for our problem called Auto MPG. Before we get stared we need to install some libraries in our Python environment. In order to do that, click on the terminal tab in PyCharm at the bottom of the window, it should have (venv) as its set environment. Run the following commands in it.

pip install pandas
pip install sklearn
pip install matplotlib

Our dataset is present in the file "auto-mpg.data", so download it inside your project directory.

Preparing the Dataset

First let's see how can we explore the dataset in Python. Observe that the values of each row in the dataset are seperated by spaces which are not fixed in length. Thus it would require us to read the file line by line in Python. Below is the source code to do the needfull.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import sklearn
from sklearn.linear_model import LinearRegression

headers = ['cylinders','displacement','horsepower','weight'
           ,'acceleration','model year','origin']
mpg = [] #Dependent variable : mpg (miles per gallon)
features = [] #Independent variables: rest of them except car name

#Preparing the data
raw = open('auto-mpg.data')
for line in raw:
    i=0
    var = []
    for y in line.split():
        if (i==0):  # The first column is mpg
            mpg.append([float(y)])
        if (i<8 and i>0):  # Not including the last column i.e. "car names"
            if(y=='?'): # Handaling missing values for "horsepower" 
                y = 0   # (Setting them to 0) Later we will change them
            if(i==4):
                var.append(float(y.replace('.','')))
            else:
                var.append(float(y))
        i+=1
    features.append(var)
    
 
df_x = pd.DataFrame.from_records(features, columns = headers)
df_y = pd.DataFrame.from_records(mpg, columns = ['mpg'])

Now we have the entire datset in the variables: 'df_x' and 'df_y'. You can preview these variables and get an overall summary of them by using:

print(df_x.describe(include='all'))
print(df_y.describe(include='all'))
print(df_x.head()) # Prints the top 5 values
print(df_y.head()) # Prints the top 5 values

Now let us change the missing values, i.e. now 0, to their average values.

avg_bhp = (np.average(df_x['horsepower'])*len(df_x))/(len(df_x)-6)
df_x['horsepower'] = df_x['horsepower'].replace(0,avg_bhp)

Now let's train our first ML model in Python.

X_train, X_test, Y_train, Y_test = sklearn.model_selection.train_test_split(
        df_x,df_y,test_size=0.2,random_state = 5)
lm = LinearRegression()

lm.fit(X_train, Y_train)
pred_test = lm.predict(X_test)

print('coefficients :(all vars)')
print(lm.coef_)

print('Intercept :(all vars)')
print(lm.intercept_)

print('Accuracy Score(all vars): %s' % lm.score(X_test,Y_test))

residues = (Y_test - pred_test)

plt.scatter(X_test, Y_test,  color='black') # The actual values
plt.plot(X_test, pred_test, color='blue', linewidth=3) # The predicted values

Voila! You have successfully completed your first ML model.

License

This guide is free to use for non-commercial purposes. For any improvements please reach out to me on: bhanu93(dot)iitd(at)gmail(dot)com.