H2O à la dplyr (R) or à la tidypolars (Python)....

[coforfe]

Hi,

It would be very interesting if h2o could have a syntax to chain transformation stages based on the use of pipes, equivalent to the one that already exists in R with dplyr, recently exported to Julia with tidier.jl. Within Python there are already alternatives such as datar and tidypolars. Not to mention that for BigData, sparklyr also uses this approach.

All these approaches resemble each other, allowing the use of different languages in a simpler way. The advantages of readability, maintainability of code based on this syntax is what is leading to the emergence of these cross-language equivalences.

I believe that this syntax would give an extra advantage to h2o to be able to encapsulate transformations or pipelines that further facilitate its implementation in production as well as making easier its learning and adoption.

• https://github.com/TidierOrg/Tidier.jl
• https://github.com/markfairbanks/tidypolars
• https://github.com/pwwang/datar

Thank you very much,
Carlos.