There are two types of features, numerical feature and categorical feature.
In the categorical feature, each observation is one of a discrete set of possible values.
Note that some features may have to be handled as the categorical feature even though they are looks like numerical features.
(e.g. Country names may be being replaced to country codes.)

Most machine learning algorithms cannot handle the categorical feature directly.
So, the categorical features commonly transform to the numerical features.
MALSS automatically transform the categorical features to the numerical features using dummy variables.