Stefan Otte

Ow, that super nice of you to say. Thanks!

[albumentations](https://albumentations.readthedocs.io/en/latest/) is also a contender for faster image augmentation. In my experience IO is actually worse than a "slow pre-processing" library. SSDs and NVMes(!) help a lot.

@soroosh-rz one-hot-encode your categorical feature and turn it into time series. Let's say your cat feature has "a", "b", and "c". Turn this into three time series, one which is...

I just played with `overload` and `singledispatch`. This is mostly a note to myself, and maybe it's helpful for others. The combination of `singledispatch` and `overload` is tricky and mypy...

This would be the fix: `var_list = var_create(*args or __data.columns)`. And with some context and a doctest: ```python @singledispatch2(pd.DataFrame) def gather(__data, key, value, *args, drop_na=False, convert=False): """Reshape table by gathering...

As usual there are time constraints regarding contributions :( Also right now this issue is not well defined yet. Do we just want to use existing serialization formats? Do we...

@aarnphm thanks for the quick response. `darts` allows you to use a bunch of different time series libraries, it wraps them with a nice unifying interface and adds some convenient...

From a usability standpoint it totally makes sense to have a `opt.ask(use_named_args=True)` or `opt.ask_named()` or something along the lines. `point_asdict` can be a workaround. That said, I would assume the...

I don't think that `ask_named()` would create a weird API, it would create a more obvious one IMHO. `opt.ask(use_named_args=True)` would also be more obvious than having to use `point_asdict()`. But...

I use https://github.com/godlygeek/tabular for my tables. It's not integrated in vim-orgmode and not as feature rich as Emacs' orgmode table thingy but all I have to do is `:Tabular /|`...