issue 137/PR 145 automatically search for categorical variables #145
Conversation
Change tests for newer sklearn
| @@ -223,28 +223,87 @@ def from_pipeline(cls, pipeline: dict): | |||
| target_encoder, | |||
| is_fitted=pipeline["_is_fitted"], | |||
| ) | |||
|
|
|||
| def get_continous_and_discreate_columns( | |||
There was a problem hiding this comment.
get_continuous_and_discrete_columns(
| target_column_name :str | ||
| ) -> tuple: | ||
| """Filters out the continious and discreate varaibles out of a dataframe and returns a tuple containing lists of column names | ||
| It assumes that numerical comumns with less than or equal to 10 different values are categorical |
There was a problem hiding this comment.
continuous instead of continious
discrete instead of discrete
variables instead of varaibles
columns instead of columns
| Parameters | ||
| ---------- | ||
| df : pd.DataFrame | ||
| DataFrame that you want to divide in discreate and continous variables |
| f"""Cobra automaticaly assumes that following variables are | ||
| discrete: {discrete_vars} | ||
| continuous: {continuous_vars} | ||
| If you want to change this behaviour you can specify the discrete/continuous variables yourself with the continuous_vars and discrete_vars keywords. \nIt assumes that numerical comumns with less than or equal to 10 different values are categorical""" |
| ): | ||
| """Fit the data to the preprocessing pipeline. | ||
| If you put continious_vars and target_vars equal to `None` and give the id_col_name Cobra will guess which varaibles are continious and which are not |
There was a problem hiding this comment.
typos:
continuous, variables, continuous
| """ | ||
| if not (continuous_vars and discrete_vars): | ||
| continuous_vars, discrete_vars = self.get_continous_and_discreate_columns( |
There was a problem hiding this comment.
Typos in code can happen, but please always double-check your method names for typos, since this snowballs to the end user eventually.
get_continuous_and_discrete_columns(...)
| """Fit preprocessing pipeline and transform the data. | ||
| If you put continious_vars and target_vars equal to `None` and give the id_col_name Cobra will guess which varaibles are continious and which are not |
|
Hi @patrickleonardy, it looks like Jano reviewed this after you and Pedro merged it. Still, his comments are valuable. Can you check your work against his comments? |
|
Hi Sander, |
sandervh14
changed the title
sandervh14
changed the title
sandervh14
changed the title
Story Title
Automatically search for categorical variables #137
Changes made
get_continous_and_discreate_columnsfunction to thePreProcesserclass.fitandfit_transformto automatically callget_continous_and_discreate_columnsif bothcontinuous_vars, discrete_varsparameters are equal toNone.How does the solution address the problem
One can now set the values for
continuous_vars, discrete_varsinfitandfit_transformequal toNone. and Cobra will guess which variables are continuous and which are discontinuous.Linked issues
Resolves #137