💞 Would You Date Yourself?

The matching hypothesis in social psychology believes that we would be more likely to date someone that looks like us.

But is that really true?
Only one way to find out.

🚀 Day 1 — Turning My Curiosity into Code Until I Land a Tech Job

I decided to measure love with data. ❤️📊

🧠 The Execution

I’ve collected a total of 53 couples who are currently dating/married.

Then, used DeepFace (Facenet) to extract facial embeddings — which is basically a face’s digital fingerprint — and compared the similarity using t-tests, Mann-Whitney U, and permutation tests to confirm significance and ensure data integrity.

Similarity Heatmap Example (This heatmap shows the comparison between two faces, to determine how similar they are by overlapping the images.)

📈 Findings

Box Comparison (Celeb) Box Comparison (Non-Celeb)

Generally, there is not much conclusive evidence to this (p-value = 0.35 > 0.05).

But I decided to separate the couples between celebrities and non-celebrities just to test out whether there is any significant difference — and here’s what I found:

  • 🌟 Celebrity couples: much more similar than random pairs (p = 0.0025 ✅ significant)
  • 👩‍❤️‍👨 Regular couples: no real difference (p = 0.35 ❌ not significant)

In other words —
Celebrity couples do look more alike with each other when compared to normal couples.

Maybe fame makes couples style themselves alike. Or maybe, it’s just better lighting. 😉

🔗 Links

🐙 View the full code and methodology on GitHub

—

💼 I’m currently looking for software engineering roles (backend, full-stack, or automation-focused).
I’m currently based in Seattle but I’m open to relocation within the United States.
📬 Feel free to reach out at LinkedIn or sinclairlimzy@gmail.com if something I built resonates with you! Looking forward to your email!