Finit classifier is a module that implement classification on finite sets.

class finit_classifier.FinitClassifier(unseen=False)

General class of final classification.

fit(y, z)

Find a function that maximizes balanced accuray.

Parameters:

  • y (numpy.array) – Features of the training set.

  • z (numpy.array) – Labels of the training set.

Returns:

None.

Return type:

None

is_seen(y)

Check if data has been seen during training.

Parameters:

y (numpy.array) – Features to try.

Returns:

List of boolean indicatin if each data point has been seen.

Return type:

list(bool)

predict(y)

Return predicted label of the input data.

Parameters:

y (numpy.array) – Features used to predict the labels.

Returns:

List of labels.

Return type:

numpy.array

class finit_classifier.index

Index a finit set.

build_index(E)

Build an index for each element of E.

Parameters:

E (numpy.array) – List to be indexed.

Returns:

None.

Return type:

None

inv(n)

Return element of E of index n.

Parameters:

n (int) – Index of an element.

Returns:

Element of E matching index n.

Return type:

Same type as the elements of E.

class finit_classifier.indexFunction

Transform functions on indexes to functions on elements using two index classes.

build_index(phi, psi)

Build an index transformation using one index class for E and one for F.

Parameters:
Returns:

None.

Return type:

None

inv(g, e)

Transform a function on indexes to a function on elements.

Parameters:

  • g (function:int->int) – Function on the indexes.

  • e (Same as elements of E) – Element to aplied the transformed function to.

Returns:

Elements matching the index g applied to the index matching e.

Return type:

Same as elements of E

class finit_classifier.optimalBA

Finit classification to maximize balanced accuracy on indexes.

fit(x, y)

Find a function that maximizes balanced accuray.

Parameters:

  • x (numpy.array[int]) – Feature of the training set.

  • y (numpy.array[int]) – Labels of the training set.

Returns:

None.

Return type:

None

predict(x)

Return prediction of labels using x.

Parameters:

x (np.array[int]) – Feature to be classified.

Returns:

Predicted labels.

Return type:

np.array[int]