calculating expanding mean for all columns in df pandas

Question

My df looks like this:

    cat1     cat2     x1     x2     x3      x4      x5      x6   . . .
0    str     str    float  float   float  float   float   float  . . .
1    str     str    float  float   float  float   float   float  . . .
.     .       .       .      .       .       .       .       .   . . .
.     .       .       .      .       .       .       .       .   . . .

I've tried this:

df = df.groupby(['cat1','cat2']).apply(pd.expanding_mean)

but that gives me a

ValueError: could not convert string to float: 

The only strings are in the groupby. This works fine but isn't what I need:

df = df.groupby(['cat1','cat2']).mean()

Answer 1

Thanks also to other users' efforts, the following might be a solution:

df.iloc[:,2:] = df.groupby(['cat1','cat2']).transform(pd.expanding_mean)

Which preserves the first two columns.