将M x（n * N）转换为（n * M）x N数据帧

Question

I have a dataframe df1 with M rows and n*N columns. The columns are named

1.1,...,1.N,2.1,...,2.N,...n.1,...nN

The rows are indexed 1,...M .

What is an elegant way to turn this into a new dataframe df2 with n*M rows indexed 1.1,...,1.M,...,n.1,...nM and N columns named 1,...,N , such that

df2.iloc[i*M+j,:]==df1.iloc[j,i*n:(i+1)*N]

Answer 1

Use Series.str.split with DataFrame.stack and last flatten MultiIndex by f-string s:

df.columns = df.columns.str.split('.', expand=True)
df = df.stack(0)
df.index = [f'{a}.{b}' for a, b in df.index]