It's a cool project, thanks for sharing. graphic-walker focuses more on a Grammar of Graphics based graphic system, not a chart-based system.
But for performance in the browser, I am also researching ways to handle large datasets in the browser, such as duckdb-wasm.
Another project I am working on, called RATH is facing more challenging performance issues because it is designed to automate the exploration of data to find insights and causal relations by machine. It currently is using web worker and indexedDB to implement a lite cube service in the browser, but it is not enough.
Besides automated data profiling, RATH focuses more on EDA, especially good at high-dimensional cases. It can generate visualization with insights about complex relations of multi variables (for example, 6 vars).
You can find the difference of RATH in mega/semi-automation module(based on predictive interaction), and in data painter, and causal analysis as well.
But for performance in the browser, I am also researching ways to handle large datasets in the browser, such as duckdb-wasm.
Another project I am working on, called RATH is facing more challenging performance issues because it is designed to automate the exploration of data to find insights and causal relations by machine. It currently is using web worker and indexedDB to implement a lite cube service in the browser, but it is not enough.
https://github.com/kanaries/rath
Thanks for your advice, I will continue to improve the both computation and render performance in browser.