.. _fwflag_test_combined_regression:
============================
--test_combined_regression
============================
Switch
======

--test_combined_regression FEAT_TABLE [FEAT_TABLE ...]

Description
===========

Trains and tests a "combined" regression model that fits a separate sub-model per feature table (one model per :doc:`fwflag_f` table, plus one per table given here) and combines their predictions, rather than concatenating all features into a single model.

Argument and Default Value
==========================

One or more *additional* feature table names — not the same as :doc:`fwflag_f`, which supplies the first table(s). Default: ``[]``.

Details
=======

Verified end-to-end: with a 1to3gram table via ``-f`` and a Facebook-topics table passed to this switch, ran a full train/test cycle and printed per-model evaluation (R², Pearson r, Spearman rho, MSE, MAE) followed by a ``***Combined Results***`` section.

Other Switches
==============

Required Switches:

* :doc:`fwflag_d`, :doc:`fwflag_c`, :doc:`fwflag_t`
* :doc:`fwflag_f` (the primary feature table)
* :doc:`fwflag_outcome_table`, :doc:`fwflag_outcomes`

Optional Switches:

* :doc:`fwflag_combined_models`

Example Commands
================

Using the packaged ``dla_tutorial`` dataset (see :doc:`../../tutorials/tut_dla`):

.. code-block:: bash

	dlatkInterface.py -d dla_tutorial -t msgs -c user_id -f 'feat$1to3gram$msgs$user_id$0_05' \
	--outcome_table blog_outcomes --outcomes age \
	--test_combined_regression 'feat$cat_met_a30_2000_cp_w$msgs$user_id$16to16'