Pandas max value index

问题:

I have a Pandas DataFrame with a mix of screen names, tweets, fav's etc. I want find the max value of 'favcount' (which i have already done) and also return the screen name of that 'tweet'

df = pd.DataFrame()
df['timestamp'] = timestamp
df['sn'] = sn
df['text'] = text
df['favcount'] = fav_count


print df
print '------'
print df['favcount'].max()

I cant seem to find anything on this, can anyone help guide me in the right direction?

回答1:

use .argmax() to get the index of the max value. then you can use loc

df.loc[df['favcount'].idxmax(), 'sn']

edit: argmax() is now deprecated, switching for idxmax()

回答2:

I think you need idxmax - get index of max value of favcount and then select value in column sn by ix:

df = pd.DataFrame({'favcount':[1,2,3], 'sn':['a','b','c']})

print (df)
   favcount sn
0         1  a
1         2  b
2         3  c

print (df.favcount.idxmax())
2

print (df.ix[df.favcount.idxmax()])
favcount    3
sn          c
Name: 2, dtype: object

print (df.ix[df.favcount.idxmax(), 'sn'])
c