Implement Machine Learning with Text using scikit-learn at PyCon 2016

Introduction:
Machine learning with text data is a powerful tool that can extract meaningful insights and predictions from unstructured text data. In this tutorial, we will explore how to use scikit-learn, a popular Python library for machine learning, to perform text classification and sentiment analysis on text data.

Dataset:
For this tutorial, we will be working with a text dataset that contains movie reviews labeled with their sentiment (positive or negative). This dataset is commonly used for sentiment analysis tasks in machine learning. You can download the dataset from the following link: https://www.cs.cornell.edu/people/pabo/movie-review-data/review_polarity.tar.gz

Importing libraries:
Before we start working with the dataset, we first need to import the necessary libraries. Start by importing scikit-learn and other Python libraries that we will use throughout the tutorial:

import numpy as np
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import MultinomialNB
from sklearn.metrics import accuracy_score, classification_report

Loading and preprocessing data:
Next, we will load the dataset into our Python environment and preprocess the data for machine learning tasks. We can use the following code snippet to load the data:

import tarfile
import os

# Extract the dataset
tar = tarfile.open("review_polarity.tar.gz", "r:gz")
tar.extractall()
tar.close()

# Load the data into memory
data_dir = "txt_sentoken"
reviews = []
labels = []

for label in ["pos", "neg"]:
    folder = os.path.join(data_dir, label)
    for file in os.listdir(folder):
        with open(os.path.join(folder, file), "r") as f:
            review = f.read()
            reviews.append(review)
            labels.append(label)

# Convert labels to binary
label_map = {"pos": 1, "neg": 0}
labels = [label_map[label] for label in labels]

In the above code snippet, we first extract the dataset into a directory called "txt_sentoken". We then load the data into memory by reading each review file and its corresponding sentiment label. The sentiment labels are converted to binary values (0 for negative and 1 for positive).

Creating feature vectors:
To perform machine learning on text data, we need to convert the text reviews into numerical feature vectors that machine learning algorithms can operate on. This process is called feature extraction. In this tutorial, we will use the bag-of-words model to convert text reviews into feature vectors using the CountVectorizer class in scikit-learn:

# Create a CountVectorizer object
vectorizer = CountVectorizer()

# Fit and transform the reviews into feature vectors
X = vectorizer.fit_transform(reviews)

# Split the data into training and test sets
X_train, X_test, y_train, y_test = train_test_split(X, labels, test_size=0.2, random_state=42)

In the code snippet above, we initialize a CountVectorizer object to convert the text reviews into a matrix of token counts. We then fit and transform the reviews into feature vectors using the fit_transform method. Finally, we split the data into training and test sets using the train_test_split function from scikit-learn.

Training a machine learning model:
With the data prepared and converted into feature vectors, we can now train a machine learning model on the text data. In this tutorial, we will use the Multinomial Naive Bayes classifier, a common algorithm for text classification tasks:

# Create a Multinomial Naive Bayes classifier
classifier = MultinomialNB()

# Train the classifier on the training data
classifier.fit(X_train, y_train)

# Make predictions on the test data
y_pred = classifier.predict(X_test)

# Evaluate the classifier
accuracy = accuracy_score(y_test, y_pred)
print("Accuracy:", accuracy)

report = classification_report(y_test, y_pred)
print("Classification Report:n", report)

In the code above, we create a Multinomial Naive Bayes classifier and train it on the training data using the fit method. We then make predictions on the test data using the predict method and evaluate the classifier’s performance using accuracy_score and classification_report functions from scikit-learn.

Conclusion:
In this tutorial, we explored how to perform machine learning with text data using scikit-learn. We walked through the process of loading and preprocessing a text dataset, converting text reviews into feature vectors, training a machine learning model, and evaluating the model’s performance. By following these steps, you can apply machine learning to your own text data and extract valuable insights and predictions.