CODEWITHSUNDEEP
pythonlistpandasindexingmulti-index
Peng He
Peng He

Reputation: 2213

pandas multi index select: How to select subset of a dataframe?

I have created a DataFrame with multi index named key_mat.It just likes

                   cnt
keys    minutes 
0000    21:01:00    1
0002    19:08:00    2
        16:07:00    1

I want to select a subset of key_mat using .loc

>>> key_mat.loc[['0002']] 
>>>
                   cnt
keys    minutes 
0002    19:08:00    2
        16:07:00    1

Now, I have a list of keys key_list = ['0000','0001','0004'...] and a list of minutes minutes_list = ['19:08:00','20:07:00',...].I want to select cnt column where keys in key_list and minutes in minutes_list. I try to use .loc , but I just find the way key_mat.loc[[('0000','21:01:00'),('0002','19:08:00')]]['cnt'] .But this can't satisfy my needs. I find a way to use query ,key_mat.query('keys in {key_list} and minutes in {minutes_list}'.format(key_list = key_list, minutes_list =minutes_list)).Is there a way to use .loc or .ix to finish my task?

Upvotes: 1

Views: 628

Answers (1)

jezrael
jezrael

Reputation: 862641

You need slicers, but first sort_index, because error:

KeyError: 'MultiIndex Slicing requires the index to be fully lexsorted tuple len (2), lexsort depth (1)'

key_list = ['0000','0001','0004']
minutes_list = ['19:08:00','21:01:00']

df.sort_index(inplace=True)

idx = pd.IndexSlice
print (df.loc[idx[key_list,minutes_list],:])
              cnt
keys minutes     
0000 21:01:00   1

Upvotes: 1

Related Questions