Standardization (Z-score) *
Description
Standardization (Z-score) refers to the process of normalizing every value in a dataset such that the mean of All the values is 0 and the standard deviation is 1.
This technique transforms the feature values to have a mean of 0 and a standard deviation of 1.
Info
- Standardization is less affected by outliers in the data than min-max scaling.
- Unlike min-max scaling, standardization does not restrict values to a specific range.
Formula
\[ Z = \frac{x - \mu}{\sigma} \]
- \(Z\) = standard score
- \(x\) = observed value
- \(\mu\) = mean of the sample
- \(\sigma\) = standard deviation of the sample
Example
from sklearn.preprocessing import StandardScaler
std_scaler = StandardScaler()
housing_num_std_scaled = std_scaler.fit_transform(housing_num)
مقدار \(\sigma\) برابر است با انحراف از معیار (standard deviation)