pandas dataframe.at for a whole series

Question

How can I convert the following code without using the loop, using pythonic pandas :

import pandas as pd

xls = pd.ExcelFile('...')
df_1 = pd.read_excel(xls, '...')
df_2 = pd.read_excel(xls, '...')

df_1_len = len(df_1)
results = np.empty(df_1_len)
for idx in range(df_1_len):
    results[idx] = df_2.at[df_1['x'][idx], df_1['y'][idx]]

df_1 looks like :

x	y
1	a
2	a
2	b
...	...

df_2 looks like :

	1	2	..
a	val(1,a)	val(2,a)	..
b	val(1,n)	val(2,a)	..
...	...	...

@adirabargil yes, it throws `TypeError: unhashable type: 'Series'` — mlisthenewcool, Mar 04 '22 at 10:58
try `tolist()` each of them (sorry i am not if front of conputer…) — adir abargil, Mar 04 '22 at 10:59
you should unstack df2, and then create multiIndex from the df1, imma llok into it on sunday if you wont be able to figure out how. — adir abargil, Mar 04 '22 at 12:54

score 0 · Answer 1 · answered Mar 04 '22 at 11:21

0

did you try df_2.loc[df_1['x'], df_1['y']]? Maybe this answer can help you understanding the difference between at and loc.

answered Mar 04 '22 at 11:21

PlaidMode

37
6

Thanks for your answer. However it's not equivalent to the query I did. My result is a vector, yours is a matrix. – mlisthenewcool Mar 04 '22 at 11:28

pandas dataframe.at for a whole series

1 Answers1