CODEWITHSUNDEEP
pythonpandasdataframe
leealex0201
leealex0201

Reputation: 23

Accessing each entries of DataFrame and replacing them in a better way?

This question is an additional question of my previous question that I posted. What I would like to do is to replace DataFrame's string value to its first initial string. For example,

s = pd.DataFrame({'A':['S12','S1','E53',np.NaN], 'B':[1,2,3,4]})

s.A.fillna('P', inplace=True)

This will give me a Dataframe 

   A  B
0 S12 1
1 S1  2
2 E53 3
3 P   4

But then, I would like to change the string values of column 'A' to ['S', 'S', 'E', 'P'], which is their first character. What I did is following,

for i, row in s.iterrows():
    if len(row['A']) > 1:
        s['A'][i] = row['A'][0]

and I got this warning.

/anaconda/lib/python2.7/site-packages/ipykernel/__main__.py:3:         SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: http://pandas.pydata.org/pandas-    docs/stable/indexing.html#indexing-view-versus-copy
  app.launch_new_instance()
/anaconda/lib/python2.7/site-packages/ipykernel/__main__.py:7: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy

I understand that this is a non-preferred way, but what exactly I am doing inefficiently and what would be the preferred way? Is it possible to do it without converting them to numpy array?

Thank you!

Upvotes: 1

Views: 55

Answers (3)

Scott Boston
Scott Boston

Reputation: 153460

You are getting the 'SettingwithCopyWarning' because of the way you are assigning back to your dataframe. If you wish to use your "non-preferred" way, then you can avoid this warning message by using .loc:

for i, row in s.iterrows():
    if len(row['A']) > 1:
        s.loc[i,'A'] = row['A'][0]

Output:

   A  B
0  S  1
1  S  2
2  E  3
3  P  4

NOTE: You can get more info on index chaining in the Pandas docs here. Also, there are some good SO posts on 'SettingWithCopyWarning'.

Upvotes: 0

Sahil Dahiya
Sahil Dahiya

Reputation: 721

You can use apply method to trim text in each row. Also to not get SettingWithCopyWarning: warning you have to use .loc and copy()

s = s.copy()

s.loc[:,"A"] = s.A.apply(lambda x : x[0])

print(s)

    A   B
0   S   1
1   S   2
2   E   3
3   P   4

Upvotes: 0

jezrael
jezrael

Reputation: 862851

You can use fillna with indexing with str by str[0]:

s['A'] = s['A'].fillna('P').str[0]
print (s)
   A  B
0  S  1
1  S  2
2  E  3
3  P  4

Upvotes: 1

Related Questions