Python for NLP: Sentiment Analysis with Scikit-Learn
This is the fifth article in the series of articles on NLP for Python. In my previous article, I explained how Python’s spaCy library can be used to perform parts of speech tagging and named entity recognition. In this article, I will demonstrate how to do sentiment analysis using Twitter data using the Scikit-Learn library.
Sentiment analysis refers to analyzing an opinion or feelings about something using data like text or images, regarding almost anything. Sentiment analysis helps companies in their decision-making process. For instance, if public sentiment towards a product is not so good, a company may try to modify the product or stop the production altogether in order to avoid any losses.
There are many sources of public sentiment e.g. public interviews, opinion polls, surveys, etc. However, with more and more people joining social media platforms, websites like Facebook and Twitter can be parsed for public sentiment.
In this article, we will see how we can perform sentiment analysis of text data.
Problem Definition
Given tweets about six US airlines, the task is to predict whether a tweet contains positive, negative, or neutral sentiment about the airline. This is a typical supervised learning task