Вы можете использовать:
df = pd.DataFrame({'A':[0,1,5,5,8,9,6],'B':[5,8,0,9,5,6,3]})
splitt=np.array_split(df,3)
L = {k: v for i, x in enumerate(splitt, 1)
for k, v in {f'mean_A_{i}': x['A'].mean(), f'std_B_{i}': np.std(x['B'])}.items()}
df1 = pd.DataFrame([L])
print (df1)
mean_A_1 std_B_1 mean_A_2 std_B_2 mean_A_3 std_B_3
0 2.0 3.299832 6.5 2.0 7.5 1.5
Другое решение с l oop:
L = []
splitt=np.array_split(df,3)
for i, x in enumerate(splitt, 1):
d = {f'mean_A_{i}': x['A'].mean(),
f'std_B_{i}': np.std(x['B'])}
L.append(pd.Series(d))
df1 = pd.concat(L).to_frame().T
print (df1)
mean_A_1 std_B_1 mean_A_2 std_B_2 mean_A_3 std_B_3
0 2.0 3.299832 6.5 2.0 7.5 1.5