StandardScaler

StandardScaler makes data mean 0 and standard deviation 1.

StandardScaler

StandardScaler transforms each feature with z = (x - mean) / std; as a result each column has mean 0 and standard deviation 1:

from sklearn.preprocessing import StandardScaler
sc = StandardScaler()
Z = sc.fit_transform(X)
print(Z.mean(axis=0).round(3), Z.std(axis=0))   # [0. 0.] [1. 1.]
print(sc.mean_, sc.scale_)                      # learned values

The values are unbounded; an outlier can get a large z-score like 4-5.

inverse_transform

Scaled values return to the original unit with inverse_transform. It is used to make the prediction of a model trained on a scaled target readable:

Z = sc.transform([[35, 5000]])
print(sc.inverse_transform(Z))   # [[  35. 5000.]]

sc.mean_ and sc.scale_ are learned in training; transform uses the same values on every call.

Exercise Standardise

Transform X with StandardScaler and print the result with np.round(..., 3). Then print scale_ with np.round(..., 1).

Show example solution

import numpy as np
from sklearn.preprocessing import StandardScaler, MinMaxScaler

X = np.array([[25, 3000], [35, 5000], [45, 10000]], dtype=float)

sc = StandardScaler()
Z = sc.fit_transform(X)
print(np.round(Z, 3))
print(np.round(sc.scale_, 1))

Test yourself

For a feature with mean 50 and std 10, what is the z-score of 70?
- 20
- -2.0
- 2.0
- 0.7
Show answer

2.0 — (70 - 50) / 10 = 2.
Turn the scaled values back into the original unit.
```
X_geri = sc.____(Z)
```
Show answer
```
X_geri = sc.inverse_transform(Z)
```
Match the attributes.

mean_

Learned means
scale_

Learned stds
fit_transform

Learn and transform
inverse_transform

Transform back

Next lesson: MinMaxScaler

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies