Consider an NFT marketplace, such as MagicEden. Can descriptions of the collections and other data be analyzed to discover insights? This was the inspiration to develop EdenEngine, an analytics engine for MagicEden API results.

What it does

  • EdenEngine is an analytics engine for MagicEden NFT Collections data, obtained through the MagicEden API.
  • The engine reads collections.json file generated by the Collections end-point, runs a text analytics workflow, and generates results and processed MS Excel files, that can be analyzed further.

How I built it

The engine is implemented as a KNIME workflow, using the KNIME Analytics Platform.

Challenges I ran into

  • Modeling the analytics workflow in KNIME had its challenges, including scalability. Since this was my biggest project with KNIME so far, it took me some time to put everything together.
  • Time required for modeling did not leave time for new analytics modules for analyzing the collections data further or for analyzing data from other end-points.
  • I also wish I had more time to deploy the engine on the cloud through KNIME Server and for other improvements, which are listed under the github page of the project.

Accomplishments that I am proud of

  • I have developed a working KNIME workflow, that can run independently of the number of collections and generate correct results, with almost full automation.
  • The engine is scalable up to analyzing 500 collections on a desktop and can also handle larger datasets, especially if deployed on the KNIME Server.

What I learned

  • How to create a tabular dataset and a text collection from collections.json file.
  • How to model and analyze the Collections data from MagicEden.

What's next for EdenEngine

Future work on EdenEngine is envisioned as follows:

  1. Deploying the app on the web, through KNIME Server.
  2. Developing an intuitive user interface, where the user can make simple selections and specify simple parameters, which feed into the analytics engine.
  3. Creating many more modules within the engine, so that different types of data from the API are analyzed and result files are generated for them. For example, there can be analysis of NFT listings within the collections and the activities for them; there can be analysis of tokens, wallets, and much more.
  4. Going beyond descriptive analytics (term frequencies, MDS, topic analysis) and implementing predictive and even prescriptive analytics.
  5. Using Python-related nodes within KNIME to conduct many of the visual analytics, as well as further analyses, within KNIME.
  6. Developing analytics tools/dashboards for posterior analysis of the generated results.
  7. Using image processing techniques to derive features from the contents of listed NFTs, and analyze those features to come up with insights.

Built With

  • knime
  • magicedenapi
Share this project: