In numpy
, some digits "finishing by 5
" are rounding up and down. For instance:
import numpy as np
np.array([0.025, 0.015]).round(2)
# array([0.02, 0.02])
Here 0.015
rounds up to 0.02
which is what I expect, but 0.025
rounds down to 0.02
instead of 0.03
, why so ?
Is it possible to fix that ? (Except using an epsilon for comparing float values)