Greg Park


A data-driven comparison of 11 cheesesteaks in Philadelphia

May 05, 2016

It took me three years of living in Philadelphia before I tried a cheesesteak. So when I finally decided to start, I figured I should go all out. I tried 11 different cheesesteak places (some of them multiple times) in about two months, taking detailed notes along the way.

To keep comparisons fair, I always placed the same order: whiz with onions. I was skeptical at first, but I am a believer now. I did sample (and enjoy) other styles, but after much reflection, I think whiz is the way to go.

I still have many more places to try, but so far, these are my rankings:

  1. Jimmy G’s Steaks on Broad, just outside of Fairmount
  2. Sonny’s Famous Cheesesteaks in Old City
  3. Dalessandro’s Steaks in Roxborough
  4. Lazos Pizza and Grill in Fairmount/Brewerytown
  5. Ishkabibble’s II on South Street
  6. Del Rossi’s Cheesesteak Company in Northern Liberties
  7. Steve’s Prince of Steaks in Center City
  8. Tony Luke’s in South Philly
  9. Jim’s Steaks on South Street
  10. Geno’s Steaks in South Philly
  11. Pat’s King of Steaks in South Philly

I actually did like all of them, just some more than others.

Like a good researcher, I wanted to compare cheesesteaks on a common scale and create some standardized descriptions, so I also developed a rating system after some intense pilot research. For each of the main components (bread/roll, steak, whiz, and onions), I rated several attributes on Likert scales and recorded some other basic features, like price and size.

So now that I have a nice little cheesesteak dataset, why not graph it? I used d3 to create an interactive visualization, comparing prices, sizes, and 20 other attributes of every glorious cheesesteak. You can also compare any individual cheesesteak to the average values of all cheesesteaks, to see what makes some of them so special.

EDIT: After migrating this blog, I’ve removed the data visualization related to this original post, but feel free to analyze the fthe original dataset however you’d like!