CODEWITHSUNDEEP
pythonpandas
Zen
Zen

Reputation: 13

Return Column name for specified value in pandas

My question is extension of this question. Although the question asked are same, answers are only applicable to very specific situation.

Assuming same data set, I want to add column with column name containing specific value

Date               col1    col2    col3    col4    col5    col6    col7 
01/01/2016 00:00   37.04   36.57   35.77   37.56   36.79   35.90   38.15

However, I have large number of rows, and the value I want to find is neither max nor min (say i want to col name with value 36.79).

Upvotes: 1

Views: 67

Answers (1)

jezrael
jezrael

Reputation: 862481

You can compare all values of DataFrame and test at least one True by DataFrame.any - then mask is used for filter columns names:

c = df.columns[(df == 36.79).any()].tolist()

But because working with floats and possible some accuracy problems is better use numpy.isclose with select only numeric columns:

#if possible some non numeric columns
df1 = df.select_dtypes(np.number)
c = df1.columns[np.any(np.isclose(df1, 36.79), axis=0)].tolist()

#if all columns are numeric
c = df.columns[np.any(np.isclose(df, 36.79), axis=0)].tolist()

Sample:

print (df)
               Date   col1   col2   col3   col4   col5  col6   col7
0  01/01/2016 00:00  37.04  36.57  35.77  37.56  36.79  35.9  38.15
1  01/02/2016 00:00  37.04  36.57  35.77  37.56  36.79  35.9  36.79 <-last value changed

df1 = df.select_dtypes(np.number)
c = df1.columns[np.any(np.isclose(df1, 36.79), axis=0)].tolist()
print (c)
['col5', 'col7']

Last if want first matched value if exist is possible use next with iter for possible pass default value if empty list (no value match):

print (next(iter(c), 'no match'))
col5

df1 = df.select_dtypes(np.number)
c = df1.columns[np.any(np.isclose(df1, 100), axis=0)].tolist()
print (c)
[]

print (next(iter(c), 'no match'))
no match

Upvotes: 1

Related Questions