Feature transformation alters features so that they're in the required form. It also reduces the effect of outliers, handles skewed data, and makes the model more robust. The following list shows the different kinds of feature transformation:
- Log transformation is the most common mathematical transformation used to transform skewed data into a normal distribution. Before applying the log transform, ensure that all the data values only contain positive values; otherwise, this will throw an exception or error message.
- Square and cube transformation has a moderate effect on distribution shape. It can be used to reduce left skewness.
- Square and cube root transformation has a fairly strong transformation effect on the distribution shape but it is weaker than logarithms. It can be applied to right-skewed data.
- Discretization can also be used to transform a numeric column or attribute. For example, the age of a group of candidates can be grouped into...
