Skip to content
Advertisement

Pandas find the maximum num from substring column

I’ve a dataframe look like this

0       1                                              2                                       3                    
0   {'Emotion': 'female_angry', 'Score': '90.0%'}   {'Emotion': 'female_disgust', 'Score': '0.0%'}  {'Emotion': 'female_fear', 'Score': '0.0%'}
1   {'Emotion': 'female_angry', 'Score': '0.0%'}    {'Emotion': 'female_disgust', 'Score': '0.0%'}  {'Emotion': 'female_fear', 'Score': '80.0%'}    
2   {'Emotion': 'female_angry', 'Score': '0.1%'}    {'Emotion': 'female_disgust', 'Score': '99.0%'} {'Emotion': 'female_fear', 'Score': '4.6%'} 

I want to make a separate column based on highest score values.

Like so

       Emotion

0      'female_angry'  

1      'female_fear'

2      'female_disgust'

I’ve went through many ref but I can’t relate with my problem. Any suggestions?

Advertisement

Answer

You can use pandas.apply with axis=1 for iterate over each row:

df_new = df.apply(lambda row: max([tuple(dct.values()) for dct in row], 
                                  key= lambda x: x[1]
                                 )[0], axis=1).to_frame(name = 'Emotion')
print(df_new)

Output:

          Emotion
0    female_angry
1     female_fear
2  female_disgust

Explanation:

>>> df.apply(lambda row: [tuple(dct.values()) for dct in row], axis=1)
# [('female_angry', '90.0%'), ('female_disgust', '0.0%'), ('female_fear', '0.0%')]
# [('female_angry', '0.0%'), ('female_disgust', '0.0%'), ('female_fear', '80.0%')]
# [('female_angry', '0.1%'), ('female_disgust', '99.0%'), ('female_fear', '4.6%')]

>>> max([('female_angry', '90.0%'), ('female_disgust', '0.0%'), ('female_fear', '0.0%')],
   key=lambda x : x[1])
# ('female_angry', '90.0%')

>>> ('female_angry', '90.0%')[0]
# 'female_angry'
User contributions licensed under: CC BY-SA
1 People found this is helpful
Advertisement