从 pandas dataframe 中选择值，基于另一个 dataframe 中具有最小值/最大值的列

Question

I have two dataframes with different quarters of the year as columns and particular locations as rows:

Temperature:

	q_1	q_2	q_3	q_4
A	10	50	0	5
B	6	0	30	1
C	60	2	9	16

Precipitation

	q_1	q_2	q_3	q_4
A	18	1	0	7
B	6	13	12	3
C	3	20	4	0

I am trying to create another dataframe whose columns are populated by the temperatures for the wettest/driest quarter and the precipitation of the warmest/coolest quarters for each location:

DF_new:

	temp_wettest	temp_driest	precip_warmest	precip_coolest
A	10	0	1	0
B	0	1	12	13
C	2	16	3	20

I have been trying to use idxmax:

temp = pd.DataFrame({'q_1' : [10,6,60],
                     'q_2' : [50,0,2],
                     'q_3' : [0,30,9],
                     'q_4' : [5,1,16]},index=['A','B','C'])
prec = pd.DataFrame({'q_1' : [18,6,3],
                     'q_2' : [1,13,20],
                     'q_3' : [0,12,4],
                     'q_4' : [7,3,0]},index=['A','B','C'])

DF_new = pd.DataFrame({'temp_wettest': temp[prec.idxmax(axis=1)],
                       'temp_driest' : temp[prec.idxmin(axis=1)],
                       'precip_warmest': prec[temp.idxmax(axis=1)],
                       'precip_coolest': prec[temp.idxmin(axis=1)]},index=['A','B','C'])

<OUT>
    temp_wettest    temp_driest precip_warmest  precip_coolest
A      (q, _, 1)      (q, _, 3)      (q, _, 2)       (q, _, 3)
B      (q, _, 2)      (q, _, 4)      (q, _, 3)       (q, _, 2)
C      (q, _, 2)      (q, _, 4)      (q, _, 1)       (q, _, 2)

I get why idxmax isn't working (it's just passing in a list of column names), but I'm not sure how to get the actual values into the new dataframe. I've also tried using pd.apply(), but I'm unsure of what function to use.

Thanks

Answer 1

If you have Pandas version < 1.2.0, try lookup :

DF_new = pd.DataFrame({'temp_wettest': temp.lookup(prec.index, prec.idxmax(axis=1)),
                       'temp_driest' : temp.lookup(prec.index, prec.idxmin(axis=1)),
                       'precip_warmest': prec.lookup(temp.index, temp.idxmax(1)),
                       'precip_coolest': prec.lookup(temp.index, temp.idxmin(1))
                      })

Output:

   temp_wettest  temp_driest  precip_warmest  precip_coolest
0            10            0               1               0
1             0            1              12              13
2             2           16               3              20

Answer 2

I have two dataframes with different quarters of the year as columns and particular locations as rows:

Temperature:

	q_1	q_2	q_3	q_4
A	10	50	0	5
B	6	0	30	1
C	60	2	9	16

Precipitation

	q_1	q_2	q_3	q_4
A	18	1	0	7
B	6	13	12	3
C	3	20	4	0

I am trying to create another dataframe whose columns are populated by the temperatures for the wettest/driest quarter and the precipitation of the warmest/coolest quarters for each location:

DF_new:

	temp_wettest	temp_driest	precip_warmest	precip_coolest
A	10	0	1	0
B	0	1	12	13
C	2	16	3	20

I have been trying to use idxmax:

temp = pd.DataFrame({'q_1' : [10,6,60],
                     'q_2' : [50,0,2],
                     'q_3' : [0,30,9],
                     'q_4' : [5,1,16]},index=['A','B','C'])
prec = pd.DataFrame({'q_1' : [18,6,3],
                     'q_2' : [1,13,20],
                     'q_3' : [0,12,4],
                     'q_4' : [7,3,0]},index=['A','B','C'])

DF_new = pd.DataFrame({'temp_wettest': temp[prec.idxmax(axis=1)],
                       'temp_driest' : temp[prec.idxmin(axis=1)],
                       'precip_warmest': prec[temp.idxmax(axis=1)],
                       'precip_coolest': prec[temp.idxmin(axis=1)]},index=['A','B','C'])

<OUT>
    temp_wettest    temp_driest precip_warmest  precip_coolest
A      (q, _, 1)      (q, _, 3)      (q, _, 2)       (q, _, 3)
B      (q, _, 2)      (q, _, 4)      (q, _, 3)       (q, _, 2)
C      (q, _, 2)      (q, _, 4)      (q, _, 1)       (q, _, 2)

I get why idxmax isn't working (it's just passing in a list of column names), but I'm not sure how to get the actual values into the new dataframe. I've also tried using pd.apply(), but I'm unsure of what function to use.

Thanks

Answer 3

I dont know the best alternative for lookup , but this might work.

DF_new = pd.DataFrame({'temp_wettest': temp.stack().loc[list(map(tuple,prec.idxmax(axis=1).reset_index().to_numpy()))].tolist(),
                       'temp_driest' : temp.stack().loc[list(map(tuple,prec.idxmin(axis=1).reset_index().to_numpy()))].tolist(),
                       'precip_warmest': prec.stack().loc[list(map(tuple,temp.idxmax(axis=1).reset_index().to_numpy()))].tolist(),
                       'precip_coolest': prec.stack().loc[list(map(tuple,temp.idxmin(axis=1).reset_index().to_numpy()))].tolist()})