.. index:: Core Operations; Projection
.. _core-operation-projection:

Projection
==================================
Projection reduces ledger to rows with distinct attribute values while aggregating their values.

Structure
---------

.. code-block:: json

    {
        "_schema": "Projection_1.0",
        "columns": ["EventId"]
    }

Parameters
----------

+-------------------+----------+------------------------+------------------------+
| Parameter Name    | Required | Type                   | Description            |
+===================+==========+========================+========================+
| ``columns``       | Yes      | ``string list``        | List of column names   |
+-------------------+----------+------------------------+------------------------+

Behaviour
--------------

Records cannot be reduced by ``Trial``, ``Time``, ``Type`` of ``Value`` column.
Records with different ``Trial``, ``Time`` or ``Type`` are always distinct and cannot
be combined using projection, and ``Value`` is storing accumulated value of merged records
with identical attributes.

Output records will be reduced by the attributes that are listed in the ``columns`` list. Those
attributes should exists in the loaded data.

:ref:`node_path` is a special column name that can be used in the ``columns`` list. It stores path
of the record through the graph, therefore different records may have a different path.
If :ref:`node_path` is not listed in the ``columns`` list (default behavior) then the records path is
ignored during projection and records generated by the projection have a new node path, indicating
origination at the projection.
If :ref:`node_path` is listed in the columns list, then records with different paths are not
combined and output records preserve original paths + label of the projection.