CODEWITHSUNDEEP
pythonpandasdataframefillna
parcolau
parcolau

Reputation: 137

Pandas Dataframe forward fill upto certain condition is met

I have a dataframe like this generated from the below code: df1

import pandas as pd
import numpy as np
df = pd.DataFrame({'Start': [True, '-', '-', '-', True, '-', '-', '-', True, '-', '-', '-'], 
                   'End': ['-', '-', '-', True, '-', '-', '-', True, '-', '-', '-', True], 
                   'Value': ['-', 15, '-', '-', 109, '-', '-', '-', '-', '-', 13, '-']})
df

|       | Start |  End  | Value |
|-------|-------|-------|-------|
|     0 |  True |     - |     - |
|     1 |     - |     - |    15 |
|     2 |     - |     - |     - |
|     3 |     - |  True |     - |
|     4 |  True |     - |   109 |
|     5 |     - |     - |     - |
|     6 |     - |     - |     - |
|     7 |     - |  True |     - |
|     8 |  True |     - |     - |
|     9 |     - |     - |     - |
|    10 |     - |     - |    13 |
|    11 |     - |  True |     - |

and I would like to forward fill the 'Value' column till the point where 'End' column == True like this: df2

|       | Start |  End  | Value |
|-------|-------|-------|-------|
|     0 |  True |     - |     - |
|     1 |     - |     - |    15 |
|     2 |     - |     - |    15 |
|     3 |     - |  True |    15 |
|     4 |  True |     - |   109 |
|     5 |     - |     - |   109 |
|     6 |     - |     - |   109 |
|     7 |     - |  True |   109 |
|     8 |  True |     - |     - |
|     9 |     - |     - |     - |
|    10 |     - |     - |    13 |
|    11 |     - |  True |    13 |

Any help will be highly appreaciated!

PS. Sorry my reputation is not high enough to post images directly...

Upvotes: 4

Views: 3329

Answers (1)

rafaelc
rafaelc

Reputation: 59274

First, always avoid to have mixed types in a column. Apparently you have strings and boolean values - that's not recommended.

First, make your data frame workable: 

df = df.replace('-',np.nan).astype(float)

Then, simple groupby+ffill

s = df.start.eq(1).cumsum()
df['value'] = df.groupby(s).value.ffill()

    start  end  value
0     1.0  NaN    NaN
1     NaN  NaN   15.0
2     NaN  NaN   15.0
3     NaN  1.0   15.0
4     1.0  NaN  109.0
5     NaN  NaN  109.0
6     NaN  NaN  109.0
7     NaN  1.0  109.0
8     1.0  NaN    NaN
9     NaN  NaN    NaN
10    NaN  NaN   13.0
11    NaN  1.0   13.0

Upvotes: 10

Related Questions