Python Sentiment Analysis using TextBlob and VADER for Glassdoor Reviews

Motivation

There are so many reviews on companies that people work for, it can be hard to digest all of them. What do people really think about the companies they work for? Can we count on company ratings on Glassdoor.com? Curious about what people are saying? Me too!

So, I decided to try out web scrapping on a few glassdoor.com machine learning job listings with company reviews. Since glassdoor.com does not have an available API, I was only able to scrape a small amount of reviews before getting the ominous 403 Forbidden Error 😳 Fear not, I did manage to scrape a few reviews…

Once I got my hands on some data, I thought it would be interesting to analyze company reviews and perform sentiment analysis to understand what users were writing. I was also curious about the number of stars given to a company and how it related to the sentiment in the review.

In this post, I will discuss the following:

• Data Analysis of glassdoor.com reviews: Stars given
• Sentiment Analysis using TextBlob: Polarity & Subjectivity
• Vader Sentiment Analysis: Compound scores

Stars Given — is this all there is to it?

Before all else… a quick look at the small-ish data: it consists of about 45 unique companies with a total of 410 reviews.