gab23r

Results 34 comments of gab23r

Add assertions while piping

Thank you! I should definitely have a closer look to `pandera` But for now, I would like to have a lightweight solution. However, how do you make the first solution...

Partition_by should returns tuples as keys when partitioning on a list of a single column, like group_by does

I agree, I already raise this behavior here: #11569

Partition_by should returns tuples as keys when partitioning on a list of a single column, like group_by does

IMHO, `by='a' and by=['a']` are not equivalent, and I would prefer the output types to be align to the input types and not on the data itself. But this idea...

Why nested numba function too slow?

Maybe because `_cumsum_nb4` needs to be compiled several times ?

feat(python): Add remap method.

Is it reasonable to say that `map` could take either a Callable or a dictionary? Then the default value could be handle via `fill_null` and new column via `alias`

Enable starts_with in the string namespace to work with expressions

I am working on it, will propose a PR soon !

Add `dtypes` parameter to `DataFrame` initialization

A `schema` argument would be super useful especially to create an empty dataframe from the schema of another one This is super useful to create mwe of bugs for exemple...

Add `dtypes` parameter to `DataFrame` initialization

Cool! I didn't known that, I add an example in the doc #6225

Support more datatypes in .clip(), .clip_max() and .clip_min()

@ritchie46, we could even accept Expr for min and max argument, In this case we could maybe use when().then().otherwise() internally ? Or it will lose too much in performance compare...

Support more datatypes in .clip(), .clip_max() and .clip_min()

I am on it !