Digits Classification using Scikit-Learn

Introduction

This lab is a step-by-step tutorial on how to use classification techniques on the Digits dataset using scikit-learn. In this lab, we will load the dataset, preprocess the data, split the dataset into training and testing sets, and then use two different classification techniques (K-Nearest Neighbors and Logistic Regression) to classify the digits. Finally, we will compare the accuracy of both techniques.

VM Tips

After the VM startup is done, click the top left corner to switch to the Notebook tab to access Jupyter Notebook for practice.

Sometimes, you may need to wait a few seconds for Jupyter Notebook to finish loading. The validation of operations cannot be automated because of limitations in Jupyter Notebook.

If you face issues during learning, feel free to ask Labby. Provide feedback after the session, and we will promptly resolve the problem for you.

Skills Graph

Load the Digits dataset

We will start by loading the digits dataset using the load_digits function from scikit-learn. This function returns two arrays: X_digits containing the input data and y_digits containing the target labels.

from sklearn import datasets

X_digits, y_digits = datasets.load_digits(return_X_y=True)

Preprocess the data

We will then preprocess the data by scaling the features to a range of [0, 1] using the maximum value of the data. This can be done by dividing the input data by the maximum value of the input data.

X_digits = X_digits / X_digits.max()

Split the dataset into training and testing sets

Next, we will split the dataset into training and testing sets using scikit-learn's train_test_split function. We will use 90% of the data for training and 10% for testing.

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X_digits, y_digits, test_size=0.1, random_state=42)

Train and test the K-Nearest Neighbors classifier

We will now train a K-Nearest Neighbors (KNN) classifier using scikit-learn's KNeighborsClassifier function and test it on the testing set. We will then print the accuracy score of the classifier.

from sklearn.neighbors import KNeighborsClassifier

knn = KNeighborsClassifier()
knn.fit(X_train, y_train)
knn_score = knn.score(X_test, y_test)

print("KNN score: %f" % knn_score)

Train and test the Logistic Regression classifier

We will now train a Logistic Regression classifier using scikit-learn's LogisticRegression function and test it on the testing set. We will then print the accuracy score of the classifier.

from sklearn.linear_model import LogisticRegression

logistic = LogisticRegression(max_iter=1000)
logistic.fit(X_train, y_train)
logistic_score = logistic.score(X_test, y_test)

print("Logistic Regression score: %f" % logistic_score)

Compare the accuracy of both classifiers

Finally, we will compare the accuracy of both classifiers by printing the accuracy scores of both classifiers.

print("KNN score: %f" % knn_score)
print("Logistic Regression score: %f" % logistic_score)

Summary

In this lab, we learned how to use classification techniques on the Digits dataset using scikit-learn. We loaded the dataset, preprocessed the data, split the dataset into training and testing sets, and then trained and tested two different classifiers (K-Nearest Neighbors and Logistic Regression) on the testing set. Finally, we compared the accuracy of both classifiers.