CODEWITHSUNDEEP
pythonpandasdataframe
Rookie_123
Rookie_123

Reputation: 2017

Deriving new dataframe from a dataframe in pandas

I have a dataframe df1 that looks like this:

  Z1  Z2  Z3 Z4
   0   0   A  A
   0   B   0  0
   C   0   C  0
   D   0   D  0
   0   0   E  0

I want to make a new dataframe df2 that looks like this:

  Z1  Z2  Z3 Z4
  C   B   A  A
  D       C  
          D  
          E

For that I am declaring a blank dataframe df2 which has column Z1 Z2 Z3 Z4.
I am doing so with the help of for loop which is as follows:

periods=['Z1', 'Z2', 'Z3', 'Z4']

for i in periods:
    for j in range(0,len(df1)):
        if df1.loc[j:j,i:i]!=0:
           #I want to write such that non zero value will be written in df2
           #Like
           #df2[something]=df1.loc[j:j,i:i]

But the length of df2 is 0 as of yet, and so I can't access its cells by indexing, and I am not able to sort out how to put values into df2.

Note:This question is different from the one marked as duplicate my explanation for same is in comments

Upvotes: 0

Views: 242

Answers (3)

jpp
jpp

Reputation: 164833

This is one way.

df = pd.DataFrame({k: pd.Series(list(filter(None, df[k]))) for k in df}).fillna('')

#   Z1 Z2 Z3 ZR
# 0  C  B  A  A
# 1  D     C   
# 2        D   
# 3        E   
# 4

This may be more efficient than lambda for larger dataframes.

Upvotes: 1

cmaher
cmaher

Reputation: 5225

You can accomplish this by using pd.DataFrame.apply with pd.Series + a list comprehension:

df.apply(lambda x: pd.Series([i for i in x if i != '0']))

#     Z1   Z2 Z3   Z4
# 0    C    B  A    A
# 1    D  NaN  C  NaN
# 2  NaN  NaN  D  NaN
# 3  NaN  NaN  E  NaN

If you'd rather have empty strings thanNaN, add a call to .fillna:

df.apply(lambda x: pd.Series([i for i in x if i != '0'])).fillna('')

#   Z1 Z2 Z3 Z4
# 0  C  B  A  A
# 1  D     C   
# 2        D   
# 3        E

Upvotes: 1

BENY
BENY

Reputation: 323376

Is this what you need ? 

df.replace('0',np.nan).apply(lambda x: sorted(x,key=pd.isnull)).dropna(thresh=1).fillna('')

Out[681]: 
  Z1 Z2 Z3 Z4
0  C  B  A  A
1  D     C   
2        D   
3        E

Upvotes: 1

Related Questions