.. _funcindex:

Index of functions per column type
==================================

This is a list of the functions that can be applied on tables or columns, depending on the column type.

Table functions
-----------------
The following functions can be applies on any table, regardless of the types of columns it contains:

**Methods** (e.g. ``table.shuffle()``)

* :meth:`.CDataFrame.add_prefix`
* :meth:`.CDataFrame.add_suffix`
* :meth:`.CDataFrame.append`
* :meth:`.CDataFrame.assign`
* :meth:`.CDataFrame.describe`
* :meth:`.CDataFrame.project`
* :meth:`.CDataFrame.rename`
* :meth:`.CDataFrame.sample`
* :meth:`.CDataFrame.shuffle`
* :meth:`.CDataFrame.slice`

**Global functions** (e.g. ``cd.concat(table, table2)``):

- :func:`.concat`
- :func:`.remove_objects`

Any column
----------

The following functions can be applied on any column, regardless of its type:

**Table methods**

- :meth:`.CDataFrame.filter` [1]_

**Column methods** (e.g. ``table["column_name"].as_table()``)

* :meth:`.CSeries.as_table`
* :meth:`.CSeries.astype`
* :meth:`.CSeries.get`
* :meth:`.CSeries.if_else` [1]_
* :meth:`.CSeries.with_threshold`
* :meth:`.CSeriesColRef.as_table`
* :meth:`.CSeriesColRef.count`


Numeric column
---------------

The following functions can be applied to any numeric column.
The current column types are integer, nullable integer, integer vector and fixed point numbers.


Integer column
"""""""""""""""""

These methods apply to integer columns:

**Table methods**

- :meth:`.CDataFrame.groupby`
- :meth:`.CDataFrame.max`
- :meth:`.CDataFrame.min`
- :meth:`.CDataFrame.validate`

**Column methods**

* :meth:`.CSeriesColRef.in_range`
* :meth:`.CSeriesColRef.max`
* :meth:`.CSeriesColRef.mean`
* :meth:`.CSeriesColRef.min`
* :meth:`.CSeriesColRef.sum`
* :meth:`.CSeriesColRef.sum_in_range`
* :meth:`.CSeriesColRef.sum_squares`
* :meth:`.CSeriesColRef.var`

**Global functions**

- :func:`.cut`
- :func:`.merge`
- :func:`.series_max`
- :func:`.series_min`

**Grouping functions**

These functions require a ``GroupBy`` object (output by :meth:`.CDataFrame.groupby`) which only works for integer or date columns. Such object represents a grouped table.

.. versionadded:: 1.8

    Groupings also work for date columns

Performing the following 

**Grouping table methods**

* :meth:`.CDataFrameGroupBy.as_table`
* :meth:`.CDataFrameGroupBy.col`
* :meth:`.CDataFrameGroupBy.size`

**Grouping column methods**

For the following methods, not only must the grouping be created for an integer column, but it can only be applied to an integer column

- :meth:`.CSeriesGroupBy.agg`

Given a table ``tab`` with an integer index ``int_index``, a column of ages ``age`` and one column with names ``name``:

.. code:: python

    grouping = tab.groupby("int_index")

    # This works
    age_sum = grouping["age"].agg(cd.groupby.sum)

    # This does not
    name_sum = grouping["name"].agg(cd.groupby.sum)

Integers or fixed-points
"""""""""""""""""""""""""""""""

.. versionadded:: 1.7

    Fixed-point columns

Constructing a :ref:`logistic<logreg>` or :ref:`linear<linreg>` regression model can be done with either integer or fixed-point columns using the following methods

* :meth:`.crlearn.linear_model.LinearRegression.fit` (inherited from :class:`Ridge<.crlearn.linear_model.Ridge>`)
* :meth:`.crlearn.linear_model.Ridge.fit`
* :meth:`.crlearn.logistic_regression.LogisticRegression.fit`

Column of vectors of integers
"""""""""""""""""""""""""""""""

**Column methods**

* :meth:`.CSeries.inner`
* :meth:`.CSeries.vsum`



String column 
---------------

**Column methods**

* :meth:`.CSeries.contains`
* :meth:`.CSeries.fullmatch`
* :meth:`.CSeries.len`
* :meth:`.CSeries.lower`

.. versionadded:: 1.8

    String matching

**Global functions**

- :func:`.merge` [2]_


Date column
---------------

.. versionadded:: 1.8

    Date columns 

* :meth:`.CSeries.weekday()`
* :meth:`.CSeries.day()`
* :meth:`.CSeries.month()`
* :meth:`.CSeries.year()`

Nullable columns
-----------------

.. versionadded:: 1.8

    Any column type can be nullable instead of only integers
        
**Table methods**

- :meth:`.CDataFrame.fillna`

**Column methods**

* :meth:`.CSeries.fillna`
* :meth:`.CSeries.isna`
* :meth:`.CSeries.isnull`
* :meth:`.CSeries.notna`
* :meth:`.CSeries.notnull`


.. [1] These functions require a boolean column input but such input can be computed from a column of any type, e.g. ``tab["integer"] > 3`` or ``tab["string"] == "test"`` or simply ``tab["boolean"]``.
.. [2] Except for ``how="left"`` as that requires a :class:`.CDataFrameGroupBy` that can only exist for integer or date columns.