Closed Softdev1 closed 1 year ago
started investigating on danfo-js & pandas-js
will analyse it's compatability, comunity support etc and test out with a code implementaion
summary on danfojs-node:
this lib has its recent release on 10/2022, almost a year past.
while the stats show it's having a decent community support https://snyk.io/advisor/npm-package/danfojs-node and minor date realted compatability issues
Documentation is good
summary on pandas-js
https://snyk.io/advisor/npm-package/pandas-js
this package was last updated 6 yrs ago, so we can't consider this with this one stat, since our code has lot of scope to grow in coming time
the other alternative is to use Objects in js, but this can be a good idea if the data underneath is smaller in size.
as we are expecting a larger scraping in the future, this may lead to refactorings if we do it now.
danfo-js has vectorization in place that can handle & apply changes with relatively heavier data
i doubt you are gonna end up in an scenario where memory and objects in js is a problem
i doubt you are gonna end up in an scenario where memory and objects in js is a problem
haha, not really though! Arrays and Objects will be a good take.
[x] find an better alternative that can handle dataframes in java script
[x] In case of libraries have low community support, figure out how we can handle that internally in js/ts code
[x] whatever the technique we have followed, that should be compatible with future integrations and langchain.js
[x] make sure to update the related changes in existing code base by following up related issues