About the project

I'm Andrea, a software engineer from Italy with a background in Web Information and Data Engineering. I like data, the web and small projects that turn a curiosity into something real.

Every Monday I watch HumanSafari (Nicolò Balini) wander through supermarkets around the world. At some point he started dropping Nutella prices into his videos, almost in passing.

What if a script read every one of his videos and pulled out each Nutella price he ever mentioned?

How the first version of the dataset was built:

• Scraped the transcripts of Nicolò's videos.
• Searched them for "Nutella" with a simple regex.
• Took a window of text around each match and looked for nearby prices and weights.
• Extracted weight, local price and currency with more regexes.
• Inferred the country from the video title when possible.
• Asked an AI to label each entry: ready, missing data, ambiguous or false positive.
• Reviewed everything by hand, often jumping to the exact moment in the video.

I probably missed a few entries. Still, I got a clean, usable dataset without watching thousands of hours of footage.

More than that, a fun idea became something you can explore, compare and, maybe, add to.

If you enjoy this project even half as much as I enjoyed building it, that is already a win.

Want the details? Read how the numbers are made or the guide.