CODEWITHSUNDEEP
pythonpandas
Seb_aj
Seb_aj

Reputation: 435

python pandas right fill values

I am trying to replicate a "right fill" excel-like function which fills the values right till the next value is not null/nan/empty. This "right-fill" exercise is only to be done if the value in the immediate following row in not empty or "nan". I have the following pandas dataframe dataset. My current input table is "have". My output table is "want".

import pandas as pd
have = pd.DataFrame({ \
"0": pd.Series(["abc","1","something here"]) \
,"1": pd.Series(["","2","something here"]) \
,"2": pd.Series(["","3","something here"]) \
,"3": pd.Series(["something","1","something here"]) \
,"4": pd.Series(["","2","something here"]) \
,"5": pd.Series(["","","something here"]) \
,"6": pd.Series(["","","something here"]) \
,"7": pd.Series(["cdf","5","something here"]) \
,"8": pd.Series(["","6","something here"]) \
,"9": pd.Series(["xyz","1","something here"]) \
})

want = pd.DataFrame({ \
"0": pd.Series(["abc","1","something here"]) \
,"1": pd.Series(["abc","2","something here"]) \
,"2": pd.Series(["abc","3","something here"]) \
,"3": pd.Series(["something","1","something here"]) \
,"4": pd.Series(["something","2","something here"]) \
,"5": pd.Series(["","","something here"]) \
,"6": pd.Series(["","","something here"]) \
,"7": pd.Series(["cdf","5","something here"]) \
,"8": pd.Series(["cdf","6","something here"]) \
,"9": pd.Series(["xyz","1","something here"]) \
})

Upvotes: 3

Views: 2208

Answers (1)

piRSquared
piRSquared

Reputation: 294228

Create a boolean mask on row 2.

  • test if it's null (None or np.nan)
  • test if it's equal to an empty string ''

Assignment

  • Use loc to assign
  • replace forward fills nulls by default.
cond = have.loc[1].isnull() | have.loc[1].ne('')
have.loc[0, cond] = have.loc[0, cond].replace('', None)
have

enter image description here

If the blanks '' are white space ' ' we can use strip

cond = have.loc[1].isnull() | have.loc[1].ne('')
have.loc[0, cond] = have.loc[0, cond].str.strip().replace('', None)
have

Upvotes: 2

Related Questions