Artsts
Ethereum Price Change versus Volume Change Model Based Predictions
Project Description
Artsts - ETH Price Model Preidctions based on historical Price change vs. Volume change
This project uses the Dune Analytics API to reference the following query (https://dune.com/queries/1532819). The data is then cleaned and mutated such that it can use the powers of Tensorflow.js to create a sequential model. The data is converted to tensors and used for machine learning. Training using the Mean Squared Error Loss Function with a batch size of 32 and runs 50 total epochs. The model is then tested with normalized data and then graphed on the scatter plot as shown in the UI.
Powered by Dune Analytics and Tensorflow.js
How it's Made
This project uses Tensorflow.js and Dune Analytics to first query for the data, then use shuffled batches to train and test the model. It is built using NextJS, React, and Typescript. The useSWR hooks from Vercel are used to ensure that the data results from the initial query are revalidated as they should be and that we stay up to date with the latest query status.
