Polars: Blazingly Fast DataFrames in Rust and Python

Поделиться
HTML-код
  • Опубликовано: 10 дек 2024
  • НаукаНаука

Комментарии • 29

  • @JeremyChone
    @JeremyChone 2 года назад +28

    Great video. I have been playing with Polars, Python, and Rust, and it is a fantastic technology.
    Also, I think Polars' Python facade with Rust backend / clean API is a great model for other data-centric technologies that need performance and scripting facade interfaces.
    Big thanks to the Polars team for this gem.

  • @АнатолийБугаков-е9г
    @АнатолийБугаков-е9г 2 года назад +15

    Polars is just great! After trying it I won't go back to pandas.

  • @ControlTheGuh
    @ControlTheGuh Год назад +4

    Learn rust, python and rust synergies will be all over the place in data science soon. Great work with Polars

    • @jmnunezd1231
      @jmnunezd1231 Год назад

      Why do you think that? Can you elaborate more please? Thanks!

    • @ptarslatinum2806
      @ptarslatinum2806 Год назад

      Can you please provide some insight? I am thinking of learning rust and I wish to enter this field.

  • @incremental_failure
    @incremental_failure Год назад +2

    Haven't been so excited about data analysis for a while. Now comes the chore of converting pandas to polars in a 15k line project.

  • @Josh-wb7ii
    @Josh-wb7ii 2 года назад +4

    For a lot of my job its not possible to use eagerly evaluated DFs, polars is a fantastic option.

  • @saaddahmani1870
    @saaddahmani1870 2 года назад +2

    Really great crate and great video, thanks.

  • @swalsh1123
    @swalsh1123 2 года назад

    I'm going to try this out. Thanks

  • @milo4433
    @milo4433 2 года назад +2

    I wonder if this could ever be used with multiple machines similar to spark. Polars looks very fast and with the addition of distributed compute it would be very powerful. Perhaps it was not designed to allow that though.

    • @АнатолийБугаков-е9г
      @АнатолийБугаков-е9г 2 года назад

      Polars uses all logical cores of your machine in parallel. Doesn't that mean distributed compute?

    • @milo4433
      @milo4433 2 года назад +1

      @@АнатолийБугаков-е9г I am asking specifically about cross machine distribution. For example if we had multiple 8-core machines (like VMS in aws) available, can we distribute polars work across those vms instead of being limited to a single machine and its cores.

    • @АнатолийБугаков-е9г
      @АнатолийБугаков-е9г 2 года назад

      @@milo4433 aah I see the difference now

    • @matthiasdebernardini3388
      @matthiasdebernardini3388 2 года назад

      ​@@milo4433 yea you totally could but would need to do that yourself, not sure the crate supports it out of the box

    • @kopodgorski
      @kopodgorski Год назад

      Wouldn't be a load balancer be enough to make it happen?

  • @headbangingidiot
    @headbangingidiot 2 года назад +1

    This is great, can we use polars on a databricks cluster?

    • @rakeshjamdare2049
      @rakeshjamdare2049 Год назад

      Sure, it terms of python, its just another library so can use it very conveniently!!!

  • @josedejesuslopezdiaz
    @josedejesuslopezdiaz 2 года назад +1

    great API

  • @Birdsneverfly
    @Birdsneverfly Год назад

    Does polars with with scikit?

    • @jmnunezd1231
      @jmnunezd1231 Год назад +2

      I think they work most of the time. The good think in that if something fails, you can pass polars to numpy very easily

  • @samuelemmettbray
    @samuelemmettbray 2 года назад

    Super confused. I thought Polars was deprecating in favor of Arrow2?

  • @hiongun
    @hiongun Год назад

    need pandas compatible api, just get fast conversion from pandas

    • @headder7802
      @headder7802 Год назад

      It has methods to read and export to Pandas out of the books. It can also write or read Delta tables (locally or in S3).

  • @therendra
    @therendra Год назад

    Nice presentation