[np.nan, np.nan, 12.0, 13.0, 16.0, 16.0, 17.0]])
>>> variation(x, axis=1, ddof=1, nan_policy='omit')
array([1.05109361, 0.31428986, 0.146483  ])

N)