最大最小归一化
def norm(data):
data_max, data_min = np.tile(np.max(data, axis=-1)[:, np.newaxis], [1, data.shape[1]]), np.tile(np.min(data, axis=-1)[:, np.newaxis], [1, data.shape[1]])
data_norm = (data - data_min) / (data_max - data_min)
return data_norm
对应公式为:
n
o
r
m
=
d
a
t
a
−
d
a
t
a
m
i
n
d
a
t
a
m
a
x
−
d
a
t
a
m
i
n
norm=\frac{data-data_{min}}{data_{max}-data_{min}}
norm=datamax−datamindata−datamin
均值方差归一化
data_mean, data_var = np.tile(np.mean(data, axis=-1)[:, np.newaxis], [1, data.shape[1]]), np.tile(np.var(data, axis=-1)[:, np.newaxis], [1, data.shape[1]])
data_norm = (data - data_mean) / np.sqrt(data_var)
return data_norm
对应公式为:
n
o
r
m
=
d
a
t
a
−
d
a
t
a
m
e
a
n
d
a
t
a
m
i
n
norm=\frac{data-data_{mean}}{\sqrt{data_{min}}}
norm=datamindata−datamean