EdgeFrame column_mode

---

column_mode(self, data_column, weights_column=None, max_modes_returned=None)

Evaluate the weights assigned to rows.

Parameters:

data_column : unicode

Name of the column supplying the data.

weights_column : unicode (default=None)

Name of the column supplying the weights. Default is all items have weight of 1.

max_modes_returned : int32 (default=None)

Maximum number of modes returned. Default is 1.

Returns:

: dict

Dictionary containing summary statistics.

The data returned is composed of multiple components:

mode : A mode is a data element of maximum net weight.

A set of modes is returned. The empty set is returned when the sum of the weights is 0. If the number of modes is less than or equal to the parameter max_modes_returned, then all modes of the data are returned. If the number of modes is greater than the max_modes_returned parameter, only the first max_modes_returned many modes (per a canonical ordering) are returned.

weight_of_mode : Weight of a mode.

If there are no data elements of finite weight greater than 0, the weight of the mode is 0. If no weights column is given, this is the number of appearances of each mode.

total_weight : Sum of all weights in the weight column.

This is the row count if no weights are given. If no weights column is given, this is the number of rows in the table with non-zero weight.

mode_count : The number of distinct modes in the data.

In the case that the data is very multimodal, this number may exceed max_modes_returned.

Calculate the modes of a column. A mode is a data element of maximum weight. All data elements of weight less than or equal to 0 are excluded from the calculation, as are all data elements whose weight is NaN or infinite. If there are no data elements of finite weight greater than 0, no mode is returned.

Because data distributions often have multiple modes, it is possible for a set of modes to be returned. By default, only one is returned, but by setting the optional parameter max_modes_returned, a larger number of modes can be returned.

Examples

Given a frame with column ‘a’ accessed by a Frame object ‘my_frame’:

>>> import trustedanalytics as ta
>>> ta.connect()
Connected ...
>>> data = [[2],[3],[3],[5],[7],[10],[30]]
>>> schema = [('a', ta.int32)]
>>> my_frame = ta.Frame(ta.UploadRows(data, schema))
[===Job Progress===]

Inspect my_frame

>>> my_frame.inspect()
[#]  a
=======
[0]   2
[1]   3
[2]   3
[3]   5
[4]   7
[5]  10
[6]  30

Compute and return a dictionary containing summary statistics of column a:

>>> mode = my_frame.column_mode('a')
[===Job Progress===]
>>> print sorted(mode.items())
[(u'mode_count', 1), (u'modes', [3]), (u'total_weight', 7.0), (u'weight_of_mode', 2.0)]

Given a frame with column ‘a’ and column ‘w’ as weights accessed by a Frame object ‘my_frame’:

>>> data = [[2,1.7],[3,0.5],[3,1.2],[5,0.8],[7,1.1],[10,0.8],[30,0.1]]
>>> schema = [('a', ta.int32), ('w', ta.float32)]
>>> my_frame = ta.Frame(ta.UploadRows(data, schema))
[===Job Progress===]

Inspect my_frame

>>> my_frame.inspect()
[#]  a   w
=======================
[0]   2   1.70000004768
[1]   3             0.5
[2]   3   1.20000004768
[3]   5  0.800000011921
[4]   7   1.10000002384
[5]  10  0.800000011921
[6]  30   0.10000000149

Compute and return dictionary containing summary statistics of column ‘a’ with weights ‘w’:

>>> mode = my_frame.column_mode('a', weights_column='w')
[===Job Progress===]
>>> print sorted(mode.items())
[(u'mode_count', 2), (u'modes', [2]), (u'total_weight', 6.200000144541264), (u'weight_of_mode', 1.7000000476837158)]