search

eltonlaw / impyute

Data imputations library to preprocess datasets with missing data
http://impyute.readthedocs.io/
MIT License
354 stars 49 forks source link

consider better handling of pandas dataframes #56

Open eltonlaw opened 5 years ago

eltonlaw commented 5 years ago

Side issue of #54

Needs more research. Maybe have a function to straightforwardly check/parse? Parse out non float columns? Return a pandas dataframe?

Would we need to add Pandas as a dependency?

gkovaig commented 4 years ago

If pandas dataframe passed in, consider:

  1. Returning pandas dataframe.
  2. Reinstate row names and column names of original dataframe passed in.
  3. Process/impute only columns that are float or can be cast as float.
  4. Ensure dataframe that is returned retains same sequence (of rows and columns).