Introduction to Feature Scaling
- It is the last step in feature transformation.
- Feature Scaling is a method to standardize the independent features present in the data to a fixed range
Methods to do Feature Scaling
- There are two methods to do feature scaling
- Standardization & Normalization
Standardization
- It is also called mean centring.
- After applying standardization the value of the mean of the column standardized will be 0 and the standard deviation of the column will be 1.
- This does not have any effect on the shape of the distribution. That is the shape of the distribution will remain the same.
Normalization
- The goal of Normalization is to change the value of numeric columns in the data set to a common scale without distorting the shape of the distribution.
- In normalization, we will change the value of the input columns such that they will always lie between [0,1]
Difference between Standardization and Normalization
- In normalization, we change the values to lie between [0,1] but we cannot guarantee the mean and standard deviation of the data on which we apply normalization. But in Standardization after performing standardization the mean and standard deviation will be o and 1 respectively
Similiarites in Normalization and Standardization
- In both the shape of the distribution which the column is following will remain the same after applying standardization and Normalization in most cases.