Class bluedata_groups

source code

Instance Methods

dictops.dict_of_lists

__init__(self, d, gr)
A 'grouper' function takes a DUID (a unique i.d.

source code

__len__(self)

source code

tuple(list(datum_tr), list(datum_tr), list(str))

split(self, n, seed=0)
Take groups until you have accumulated approximately n data.

source code

Inherited from object: __delattr__, __format__, __getattribute__, __hash__, __new__, __reduce__, __reduce_ex__, __repr__, __setattr__, __sizeof__, __str__, __subclasshook__

Class Variables
	HUGE = `1e+30`

Properties
Inherited from `object`: `__class__`

Method Details

init(self, d, gr)
(Constructor)

source code

A 'grouper' function takes a DUID (a unique i.d. string for a datum) and returns the name of the data group to which it belongs. This group name is used in constructing the training and test sets.

Parameters:

d (list(datum_c)) - a list of data

Returns: dictops.dict_of_lists

A dictionary full of lists; each list corresponds to one group.

Overrides: object.__init__

split(self, n, seed=0)

source code

Take groups until you have accumulated approximately n data. The trick is to make all classifiers as identical as possible by getting the same balance of classes in the training set. We are willing to sacrifice some data to do that.

Parameters:

n (float or int) - how many data in the testing set

Returns: tuple(list(datum_tr), list(datum_tr), list(str))

(testing set, training set, names in training set)

Class bluedata_groups

__init__(self, d, gr) (Constructor)

split(self, n, seed=0)

init(self, d, gr)
(Constructor)