Collective Filter by Takens Theorem

FAQs, etc.

Collective Filter is a little data 🧪 experiment.

It takes input of selected NFT projects, and suggests others.

Suggestions are distilled from pure transaction data; patterns in a collective.

Disclaimer

Collective Filter is a toy, an experiment, for fun. It should not be regarded as an oracle for any good advice, whether financial or aesthetic or dietary.

How does it work?

Collective Filter uses a statistical model built from transaction data to suggest other NFT projects. It's built from a subset of data, just a snapshot, so it won't have all your favorite projects. Currently, it only contains data from Ethereum mainnet, and is based on thousands of transactions from hundreds of projects. Using this data snapshot, the model maps your selected interests to other projects. These suggested projects reflect collective activity of other collectors. Credits: Data from Etherscan, OpenSea. Statistical modeling & coding by Takens Theorem. Some delightful goodies by Bulma, math.js, Animate.css.

How do I use it?

See here for a demonstration. I do not require or use any wallet connections, though I may add this in the future. To start, click the buttons below 'Select.' You can mouseover these buttons to get a quick description. For example, 'tx' lists projects by transactions in the current data snapshot. When the left side populates with projects, you can select them with the

plus button. This immediately generates suggestions at the right. That's it! Feel free to click around.

A few other notes. You can exclude projects from your use by clicking the

red button on the bottom right of the results. You can also click on various explorers for the projects as shown in the small button icons. There are a few other details, feel free to click around and see.

Is this any good?

This is a little experiment. Collective Filter uses a snapshot of NFT transaction data to build a statistical model. I did little to no direct curation of the output. Sometimes Collective Filter may work quite well, sometimes results might surprise. Importantly I welcome feedback here! Warning: Creators and collectors may be disappointed by the output. But in blockchain, connections across favorite projects are subject to data trends not under our control. Sometimes the model will err. Other times, it will surprise with an unwelcome reality.

Can I save my recommendations?

I have implemented one single save feature: You can click the save checkbox in ⚙ options and opt-in to store your favorites list in localStorage. So when you close this window or refresh the page, Collective Filter will hold onto your favorites. However everything else is reset, including all selected projects, suggestions and exclusions. I welcome feature requests if you have another preference for this.

What about next features?

I have many updates already coded up. These include NFT projects on other chains, different statistical models and more. If you have favored improvements or suggestions, I very much welcome them here. If you'd like to encourage me to keep going feel free to donate here or pick up a primary item from my other creative projects.

Are you advertising or boosting?

In short, no, not at all. The model uses raw data and I do not highlight or transform the data with any specific goal other than to express the underlying collective trends. An important caveat though: Any model is based on a sample of data curated with projects I've either collected or found interesting over the years. For this reason, some projects may sometimes appear at the top more than others. This was not for the purpose of boosting, but reveals the creator's own interests. An important disclaimer to wrap up:

This project has no mint.

It's just curious data.

Update history
v0.5 - 3/29/25 - Updated all curated models
v0.41 - 3/10/25 - Minor debugging (count badge, lucky sort)
v0.4 - 3/10/25 - Added wallet preloading
v0.33 - 3/07/25 - Fixed OS links for OS2 constraints
v0.32 - 3/02/25 - Updated all curated models
v0.31 - 2/09/25 - Stacked selection & suggestion; fine tunings
v0.3 - 1/19/25 - Added Base model and Base + mainnet model
v0.2 - 12/16/24 - JPG model + fixes
v0.1 - 11/27/24 - Release mainnet curated model

Collective Filter 🔍 Try FAQs