Quick Data Analysis on Tinder

David Hsu
David Hsu
Nov 16, 2017 · 5 min read

In a previous article, I did some analysis on Coffee Meets Bagel (CMB). This time, I decided to do a quick analysis on the population of Tinder in my area of Seattle and within the general range of my age of 34. I’ve been using the app for some months now and I figured I might as well do something productive with it since there was nothing happening.

In total, I collected 1441 profiles.

Technology (or lack of technology) Used

Unlike CMB where they had a relatively uniform profile and predictable data to gather, Tinder has none of that. There’s also no history so you can’t automate the process through previously seen profiles. Instead, the only things predictable about it are at least one photo and a name. Age is listed most of the time, but besides that, there’s no consistent information presented in the app. I could’ve done a profile analysis, but the many of the profiles were either blank or composed mostly of emojis.

Instead, I did a manual pass over all the profiles since I was using it anyhow and jotted some information on Google sheets. Then I converted that to a usable Excel file and ran it through my usual Jupyter notebook using Python, Pandas, Seaborn, and Matplotlib to generate the graphs below. I saw probably more than 3000 profiles, but only recorded 1441 after noticing trends and wondering how often they actually occur.

Features Recorded

As Tinder is mostly photo oriented, I decided to focus on only a few features so I wouldn’t have to write too much:

  • Age (if present)
  • Race (in categories of White, Black/African-American, Asian, Hispanic based on observation or if specifically stated in the profile)
  • Height (in the rare case they listed it in the profile)
  • Number of photos in a profile
  • Number of photos where half their face was covered with sunglasses
  • Number of photos with pets in them (either with the person or by themselves)

After I logged a lot of the profiles, I realized I should’ve added a few more features:

  • How many photos had their face visible
  • How many photos had 2+ people present where you can’t tell who the person is (or as my friend put it, “are they the only one in the goddamn photo”)
  • How many photos were nothing but random scenery

Seriously, there’s a significant amount of profiles where you can’t see their face because they’re either wearing sunglasses, or if they weren’t, they were a tiny spec on a mountain side backlit by the sun and in deep shadow. However, enough about the general observations. Let’s look at some recorded facts.

Age Range

The age is limited to 27–39 as that was my age limit on the app. Something I didn’t consider and only noticed when I graphed out the data was the linear decline in people using this app as they became older.


From the data, the people using this in Seattle are predominantly White, making up 79% of the profiles seen. This is based mostly on observation or if the profile specifically stated what race they were.


Sometimes, the profile mentions their height and usually a statement that the person has to be taller than them. There’s not many profiles who give this information (0.3%), but for those that did, the most common listed heights was 5'7", 5'8", and 5'10".

Photo Analysis

Total Photos

In terms of the quantity of photos, the majority of profiles (59%) have the max limit of 6 photos. The quality of the photos, however, are not always the best.


For the amount of sunglasses photos I kept seeing, it seems like the majority leave the sunglasses behind when taking portraits. I classified sunglasses photos as any photo where the person was wearing sunglasses which covered the eyes. Photos where it was on their head or held with their face showing didn’t count. On average, there is a 40% chance a profile will contain at least one photo with the girl wearing sunglasses and a 16% chance a profile will contain at least two sunglasses photos.

There were 10 profiles (0.69%) where all photos had the person wearing sunglasses and 38 profiles (2.64%) had at least half of the photos with the person wearing sunglasses.

Does the race make a difference?

When accounting for race:

  • If the person was White, they had a 42% chance of wearing sunglasses in at least one photo and 17% chance of sunglasses in at least two photos.
  • If the person was Asian, they had a 37% chance of at least one sunglasses photo and 14% chance of at least two sunglasses photos.
  • If the person was Black, they had a 32% chance of at least one sunglasses photo and 11% chance of at least two sunglasses photos.


I thought pets might be more common, but they made up only 18% of all profile photos where at least one photo contained a pet.

Future Work

If I was to do this again, I would write a script similar to the one I wrote for CMB that would scroll through the photos and run a face detection algorithm on the photos to detect how many people were in a photo and if a person was present. It wouldn’t be 100% accurate, but it could go through many more photos and faster than doing it manually. I’m not sure if there’s a sunglasses detection algorithm.


Based on these graphs, the conclusion is if you’re a white male 5'8" or taller living in the Seattle area, Tinder will be great for you. I expect the demographic on Tinder will vary from city to city, but at least in Seattle, for all other males, the app will be mostly useless.

David Hsu

Written by

David Hsu

Photographer, Designer, Engineer, Video Game Programmer, and all around random craftsman