DEV Community

Arindam Dawn
Arindam Dawn

Posted on • Originally published at tabandspace.com

30 Days of Python 👨‍💻 - Day 28- ML & Data Science II

#python #datascience #machinelearning #challenge

Today I explored the Scikit-Learn library and created a notebook project to go over some of the basics and try creating a machine learning model. Scikit-Learn is a vast library and it takes a lot of practice and exploration to get a grasp of it. I followed some tutorials and articles to try building a simple classifier model just to figure out how it works. It looked a bit intimidating to me but I decided to create a basic workflow in a Jupyter Notebook so that I can use it as a reference when I decide to dive deep into the ML and Data Science Domain.

Scikit-Learn is a popular Python library for Machine Learning. Scikit-Learn can process data provided to it and create machine learning models to learn patterns within the data and makes predictions using its tools.

Why Scikit-learn?

  • Built on top of numpy and matplotlib libraries
  • Has tons of built-in machine learning models
  • Lot of methods to evaluate machine learning models
  • Easy to understand and well-designed API

Usually, Machine Learning can be a bit overwhelming as it involves complex algorithms and statistics to analyze data. Scikit-learn abstracts this complexity and makes it easy to build models and train them without having to know much about mathematics and statistics.

Here is the Notebook I created today. The link to the Github repository is https://github.com/arindamdawn/jupyter_notebooks

Basics of the scikit-learn library

This notebook covers some of the basics of the amazing scikit-learn python library. Some of the important use cases of the library have been listed in this notebook which can be used as a cheat-sheet for reference.

Some of the topics covered are:

  1. Getting the data ready
  2. Selecting the appropriate algorithm/estimator for the specific problem
  3. Fit the model/algorithm to use it to make predictions on the data
  4. Evaluting a model
  5. Improve a model
  6. Saving a loading a trained model

Getting the data ready

The data used for this project will be the heart disease data set available from https://www.kaggle.com/ronitf/heart-disease-uci

import pandas as pd
import numpy as np
heart_disease = pd.read_csv('data/heart.csv')
heart_disease.head()
Enter fullscreen mode Exit fullscreen mode
age sex cp trestbps chol fbs restecg thalach exang oldpeak slope ca thal target
0 63 1 3 145 233 1 0 150 0 2.3 0 0 1 1
1 37 1 2 130 250 0 1 187 0 3.5 0 0 2 1
2 41 0 1 130 204 0 0 172 0 1.4 2 0 2 1
3 56 1 1 120 236 0 1 178 0 0.8 2 0 2 1
4 57 0 0 120 354 0 1 163 1 0.6 2 0 2 1

The aim is to predict based on the above data whether a patient has a heart disease or not. The target column determines the result and the other columns are called the features

# Create Features Matrix (X)
X = heart_disease.drop('target', axis=1)

# Create Labels (Y)
y = heart_disease['target']
Enter fullscreen mode Exit fullscreen mode

Choose the appropriate model/estimator for the problem

For this problem, we will be using the RandomForestClassifier model form sklearn which is a classification machine learning model.

from sklearn.ensemble import RandomForestClassifier
clf = RandomForestClassifier()
clf.get_params() # lists the hyperparameters
Enter fullscreen mode Exit fullscreen mode 
{'bootstrap': True,
 'ccp_alpha': 0.0,
 'class_weight': None,
 'criterion': 'gini',
 'max_depth': None,
 'max_features': 'auto',
 'max_leaf_nodes': None,
 'max_samples': None,
 'min_impurity_decrease': 0.0,
 'min_impurity_split': None,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'n_estimators': 100,
 'n_jobs': None,
 'oob_score': False,
 'random_state': None,
 'verbose': 0,
 'warm_start': False}
Enter fullscreen mode Exit fullscreen mode

Fit the model to the training data

In this step the model is split into training and testing data

# fit the model to data
from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X,y, test_size=0.2) 
# Means 20% of the data will be used as testing data
Enter fullscreen mode Exit fullscreen mode 
clf.fit(X_train, y_train);
Enter fullscreen mode Exit fullscreen mode 
# make prediction
y_label = clf.predict(np.array([0,2,3,4]))
Enter fullscreen mode Exit fullscreen mode 
y_preds = clf.predict(X_test)
y_preds
Enter fullscreen mode Exit fullscreen mode 
array([1, 1, 1, 0, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0,
       1, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 0,
       1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1], dtype=int64)
Enter fullscreen mode Exit fullscreen mode 
y_test.head()
Enter fullscreen mode Exit fullscreen mode 
72     1
116    1
107    1
262    0
162    1
Name: target, dtype: int64
Enter fullscreen mode Exit fullscreen mode

Evaluate the model

In this step the model in evaluated on the training data and test data

clf.score(X_train, y_train)
Enter fullscreen mode Exit fullscreen mode 
1.0
Enter fullscreen mode Exit fullscreen mode 
clf.score(X_test, y_test)
Enter fullscreen mode Exit fullscreen mode 
0.7704918032786885
Enter fullscreen mode Exit fullscreen mode 
from sklearn.metrics import classification_report, confusion_matrix, accuracy_score

print(classification_report(y_test, y_preds))
Enter fullscreen mode Exit fullscreen mode 
              precision    recall  f1-score   support

           0       0.77      0.71      0.74        28
           1       0.77      0.82      0.79        33

    accuracy                           0.77        61
   macro avg       0.77      0.77      0.77        61
weighted avg       0.77      0.77      0.77        61
Enter fullscreen mode Exit fullscreen mode 
print(confusion_matrix(y_test, y_preds))
Enter fullscreen mode Exit fullscreen mode 
[[20  8]
 [ 6 27]]
Enter fullscreen mode Exit fullscreen mode 
print(accuracy_score(y_test, y_preds))
Enter fullscreen mode Exit fullscreen mode 
0.7704918032786885
Enter fullscreen mode Exit fullscreen mode

Improve the model

This step involves improving the model to get more accurate results

# Try different amount of n_estimators
np.random.seed(42)
for i in range(1, 100, 10):
    print(f'Trying model with {i} estimators')
    clf = RandomForestClassifier(n_estimators=i).fit(X_train, y_train)
    print(f'Model accuracy on test set: {clf.score(X_test, y_test) * 100:.2f}%')
    print('')
Enter fullscreen mode Exit fullscreen mode 
Trying model with 1 estimators
Model accuracy on test set: 72.13%

Trying model with 11 estimators
Model accuracy on test set: 83.61%

Trying model with 21 estimators
Model accuracy on test set: 78.69%

Trying model with 31 estimators
Model accuracy on test set: 78.69%

Trying model with 41 estimators
Model accuracy on test set: 75.41%

Trying model with 51 estimators
Model accuracy on test set: 75.41%

Trying model with 61 estimators
Model accuracy on test set: 75.41%

Trying model with 71 estimators
Model accuracy on test set: 73.77%

Trying model with 81 estimators
Model accuracy on test set: 73.77%

Trying model with 91 estimators
Model accuracy on test set: 75.41%
Enter fullscreen mode Exit fullscreen mode

Save the model and load it

Will be using the pickle library from Python to save the model

import pickle

pickle.dump(clf, open('random_forest_model_1.pkl', 'wb'))

#load the model
loaded_model = pickle.load(open('random_forest_model_1.pkl','rb'))
loaded_model.score(X_test, y_test)
Enter fullscreen mode Exit fullscreen mode 
0.7540983606557377
Enter fullscreen mode Exit fullscreen mode

That's all for today. Since Machine Learning and Data Science is an ocean in itself, I decided to look into it in more details as after being more adept with its tools and concepts, share my experience as blog posts and projects. For the remaining two parts of this challenge, I would like to explore domains such as Automation Testing with Python using Selenium and create another post on a compilation of Python resources.

Have a great one!

Top comments (0)

Read next

nrysk profile image

☘️ Growing 3D grass on Your GitHub Profile

Ryosuke -

zhiheng_huang_55a68583013 profile image

PDF chat with source highlights

Zhiheng Huang -

pauq profile image

Scraping New Telegram Channels

pauq -

mikeyoung44 profile image

LLM Quantization: Balancing Accuracy and Efficiency for Real-World Deployments

Mike Young -