Skip to content

getml.data.relationship

Marks the relationship between joins in Placeholder

many_to_many module-attribute 

many_to_many: ManyToMany = 'many-to-many'

Used for one-to-many or many-to-many relationships.

When there is such a relationship, feature learning is necessary and meaningful. If you mark a join as a default relationship, but that assumption is violated for the training data, the pipeline will raise a warning.

many_to_one module-attribute 

many_to_one: ManyToONE = 'many-to-one'

Used for many-to-one relationships.

If two tables are guaranteed to be in a many-to-one relationship, then feature learning is not necessary as they can simply be joined. If a relationship is marked many-to-one, but the assumption is violated, the pipeline will raise an exception.

one_to_many module-attribute 

one_to_many: OneToMany = 'one-to-many'

Used for one-to-many or many-to-many relationships.

When there is such a relationship, feature learning is necessary and meaningful. If you mark a join as a default relationship, but that assumption is violated for the training data, the pipeline will raise a warning.

one_to_one module-attribute 

one_to_one: OneToOne = 'one-to-one'

Used for one-to-one relationships.

If two tables are guaranteed to be in a one-to-one relationship, then feature learning is not necessary as they can simply be joined. If a relationship is marked one-to-one, but the assumption is violated, the pipeline will raise an exception. If you are unsure whether you want to use many_to_one or one_to_one, user many_to_one.

propositionalization module-attribute 

propositionalization: Propositionalization = (
    "propositionalization"
)

Used for one-to-many or many-to-many relationships.

The flag means that you want a propositionalization algorithm to be used for this particular join. This is recommended when there are very many matches within the join and normal algorithms would take too long.