Classifying Text with Logistic Regression in Python | CodeSignal Learn

Classifying Text with Logistic Regression in Python | CodeSignal Learn

Lesson Overview and Goal

Hello and welcome back! Today, we are diving into the world of text classification using Logistic Regression . In some of our previous lessons, we discussed various aspects of natural language processing, such as vectorization of text data. Today, we are going to build upon those lessons and use Logistic Regression to classify messages. By the end of this lesson you will have a firm grasp on logistic regression, how to use it for text classification and evaluate the prediction accuracy.

Understanding Logistic Regression

y = \frac{1}{1 + e^{-x}}

Sign up

Join the 1M+ learners on CodeSignal

Be a part of our community of 1M+ users who develop and demonstrate their skills on CodeSignal

Building on what you've learned about loading datasets, message vectorization, and splitting data into training and testing sets, we can proceed directly with the logistic regression implementation. To create and train our logistic regression model, we use the LogisticRegression class from sklearn.linear_model . Python from sklearn.linear_model import LogisticRegression log_reg = LogisticRegression() log_reg.fit(X_train, y_train) After training, the model is ready to make predictions on the test data. Python y_pred = log_reg.predict(X_test) With these steps, you've successfully implemented logistic regression for classifying your text data.

Evaluating Logistic Regression Model

So, we have our model and it's making predictions, but how do we know if it's any good? To measure the classification performance of our model, we use metrics. There are numerous metrics available, such as accuracy, precision, recall, F1-score, but for this lesson, we will use accuracy score. Accuracy score is a commonly used metric that gives us the proportion of correct classifications our model made. For example, if our model correctly classified 84 out of 100 messages, the accuracy score is 0.84. The accuracy score can be calculated using accuracy_score from sklearn.metrics . Python from sklearn.metrics import accuracy_score print("Accuracy:", accuracy_score(y_test, y_pred)) The output will be: text Accuracy: 0.964824120603015 This indicates that our logistic regression model was able to correctly classify approximately 96.48% of the messages in the test set, which is a strong result, demonstrating the effectiveness of logistic regression for text classification tasks. However, keep in mind that "accuracy" might not always be the best metric to evaluate performance, especially for imbalanced datasets where one class far outnumber the others. In such case, evaluation metrics like precision, recall, or F1-score can be more informative.