SQLite3 1 is a popular in-memory database. It is faster than classical relational databases like PostgreSQL, but less stable under massive parallel access, and requires all contained data sets to be loaded into memory, which might fill up too much of your RAM, especially for large data sets.
As with all other databases in the unified import interface of the
getML Python API, you have to first connect to it - using
Import from SQLite3¶
By selecting an existing table from your database in the
from_db() class method, you can create a
DataFrame containing all its data.
Alternatively you can use the
read_query() methods to replace the
content of the current
DataFrame instance or
append further rows based on either a table or a specific query.
Export to SQLite3¶
You can also write your results back into the SQLite3 database. By
providing a name for the destination table in
getml.models.MultirelModel.transform(), the features generated
from your raw data will be written back. Passing it into
getml.models.MultirelModel.predict() generates predictions
of the target variables to new, unseen data and stores the result into
the corresponding table.