(
    new_df.assign(total=new_df.total.astype(int))
    .sort_values(by='total', ascending=False, inplace=False)
)

    skill                           total
0   business process improvement    996
2   microsoft access                992
3   pattern recognition             986
1   usability testing               99