CODEWITHSUNDEEP
pythonpandas
my_prog
my_prog

Reputation: 25

How to replace NaN in single column with 0 based on index

Very new to coding so please excuse my lack of knowledge
I currently have a dataframe that looks like this: 

date         A     B    C    
2006-11-01   NaN   1    NaN
2016-11-02   NaN   NaN  1
2016-11-03   1     NaN  NaN
2016-11-04   NaN   1    NaN
2016-11-05   NaN   1    NaN
2016-11-06   NaN   NaN  NaN
2016-11-07   NaN   1    NaN

What I want to do, for example, is:
replace all NaN's in column A with 0 for all dates after 2016-11-03 and be able to do this same thing for each column but with different corresponding dates.

I have tried 

for col in df:
    if col == 'A' & 'date' > '2016-11-03':
        value_1 = {'A':0}
        df = df.fillna(value=value_1)

but I received this error TypeError: unsupported operand type(s) for &: 'str' and 'str'
I'm sure this has to do with my lack of knowledge but I'm not sure how to proceed.

EDIT: what I am looking for is something like this: 

date         A     B    C    
2006-11-01   NaN   1    NaN
2016-11-02   NaN   NaN  1
2016-11-03   1     NaN  NaN
2016-11-04   0     1    NaN
2016-11-05   0     1    NaN
2016-11-06   0     NaN  NaN
2016-11-07   0     1    NaN

Upvotes: 0

Views: 48

Answers (3)

Gularss
Gularss

Reputation: 1

You can use fillna, replace empty rows with zero

if condition == 'A':
fillna(0, inplace = True)

Upvotes: 0

kartheek
kartheek

Reputation: 696

df.loc[(df.date > '2016-11-03') & (df['A'].isnull()),'A'] = 0;

Output:

A   B   C   date
0   NaN 1.0 NaN 2006-11-01
1   NaN NaN 1.0 2016-11-02
2   1.0 NaN NaN 2016-11-03
3   0.0 1.0 NaN 2016-11-04
4   0.0 1.0 NaN 2016-11-05
5   0.0 NaN NaN 2016-11-06
6   0.0 1.0 NaN 2016-11-07

Upvotes: 0

DYZ
DYZ

Reputation: 57033

The condition consists of two parts: being after a certain date and being a NaN.

condition = (df['date'] > '2016-11-03') & df['A'].isnull()

Now, choose the rows that match the condition and make the respective items in the column A equal 0:

df.loc[condition, 'A'] = 0

Upvotes: 3

Related Questions