In [6]:
%time df = pd.read_feather('test_df.feather')

Wall time: 60.8 ms
In [7]:
%time df = pd.read_csv('test_df.gzip.csv', compression='gzip')

Wall time: 3.78 s