I have a data frame with a hierarchical index in axis 1 (columns) (from a groupby.agg operation):
groupby.agg
USAF WBAN year month day s_PC s_CL
df.columns = ['_'.join(tup).rstrip('_') for tup in df.columns.values]