The initial focus was on using it for plotting in React. We wanted to make shouldComponentUpdate decisions quickly and the Immutable base of pandas-js allows for quick, cheap comparisons after querying data directly from our API. See the post on it here: “Pandas & Immutable.js” @StratoDem https://insights.stratodem.com/pandas-immutable-js-2d9bf0106...
The next step will involve performance optimizations as the aggregates will be slow at 100MB. At that scale, we do the operations server side (we've exclusively been using this client side). Thanks for raising that concern!