Yelp Dataset Challenge: An Exploration Of Steak Reviews Using NLP
Anahita Bahri
Check out my slides here!
I have many obsessions, but there’s this one thing that people associate me with, whether the person is a close friend or an acquaintance who happens to be an Instagram follower: l’entrecote, or steak frites!
As a superfan of steak frites, I’m always on the lookout for my new favorite steak spot. There are many things I take into consideration, like...
- Do I get fries?
- Is there sauce involved?
- How about wine?
- And, most importantly, how tasty is the steak?
How about those who review steakhouses on Yelp? What factors may influence a user when rating a restaurant? Why would a user give a restaurant 3 stars over 5? Does the service matter? Quality of the food? Perhaps the price?
Throughout this project, I try to uncover trends in the steak realm by exploring the most frequent words per rating (1-5 stars), uncovering various topics for the review text (overall and per rating), extracting words similar to a few of the most frequent words within the reviews, among other techniques.
As someone who is relatively new to the data science realm, I haven’t been able to answer all of these questions just yet. I have, however, most certainly begun tackling these questions.