以字符串形式返回Pandas多索引数据框中的单元格值

Question

I am trying to return a single cell value from a dataframe of about 11.000 rows and 10 columns:

df = pd.DataFrame({ 'A' : ('foo', 'foo', 'bar', 'bar'),
                    'B' : ('horse', 'car', 'horse', 'car'),
                    'C' : ('red', 'green', 'blue', 'black')})
df
     A      B      C
 0  foo  horse    red
 1  foo    car  green
 2  bar  horse   blue
 3  bar    car  black

I've tried a few versions of loc, but always get an object returned, not the actual value ('green'). For instance:

df.loc[(df['A'] == 'foo') & (df['B'] == 'car'), 'C']

returns:

1    green
Name: C, dtype: object

The result is correct, but it is returned as an object, not as a string. I also tried setting columns A and B as indices in a multi-index frame. The loc code is then easier, but same result.

So how can I get the value as a string, not as an object?

Answer 1

You are getting a series out, and you want the first object.

You can think of what you are doing as getting the pd.Series equivalent of a list of length 1: ['green'] . If it was a list, you would do list[0] .

However, pandas series subset based on index, so you can't do this. To subset without knowledge of a series index, use iloc[0] :

df.loc[(df['A'] == 'foo') & (df['B'] == 'car'), 'C'].iloc[0]

'green'

Answer 2

Use item :

pandas.Series.item

Series.item() return the first element of the underlying data as a python scalar

df.loc[(df['A'] == 'foo') & (df['B'] == 'car'), 'C'].item()

Output:

'green'