--- title: Scalers keywords: fastai sidebar: home_sidebar summary: "Utils for scaling data" description: "Utils for scaling data" nb_path: "nbs/data__scalers.ipynb" ---
{% raw %}
{% endraw %} {% raw %}
{% endraw %} {% raw %}

class Scaler[source]

Scaler(normalizer)

{% endraw %} {% raw %}

norm_scaler[source]

norm_scaler(x, mask)

{% endraw %} {% raw %}

inv_norm_scaler[source]

inv_norm_scaler(x, x_min, x_max)

{% endraw %} {% raw %}

norm1_scaler[source]

norm1_scaler(x, mask)

{% endraw %} {% raw %}

inv_norm1_scaler[source]

inv_norm1_scaler(x, x_min, x_max)

{% endraw %} {% raw %}

std_scaler[source]

std_scaler(x, mask)

{% endraw %} {% raw %}

inv_std_scaler[source]

inv_std_scaler(x, x_mean, x_std)

{% endraw %} {% raw %}

median_scaler[source]

median_scaler(x, mask)

{% endraw %} {% raw %}

inv_median_scaler[source]

inv_median_scaler(x, x_median, x_mad)

{% endraw %} {% raw %}

invariant_scaler[source]

invariant_scaler(x, mask)

{% endraw %} {% raw %}

inv_invariant_scaler[source]

inv_invariant_scaler(x, x_median, x_mad)

{% endraw %} {% raw %}
{% endraw %}

Usage Example

{% raw %}
data = 100 + 0.5*np.random.randn(1000)
print('Mean:', data.mean())
print('STD:', data.std())
Mean: 100.03536751831068
STD: 0.5036039038433381
{% endraw %}

Apply scaler

{% raw %}
scaler = Scaler(normalizer='std')
data_norm = scaler.scale(x=data, mask=np.ones(data.shape))
{% endraw %}

Check mean and std

{% raw %}
print('Mean:', data_norm.mean())
print('STD:', data_norm.std())
Mean: 2.799538378894795e-15
STD: 1.0
{% endraw %}