Package sentence :: Module ranking

Module ranking

Utility functions and classes for ranking. Can be used both in a pythonic or an object-oriented way wrapped in Ranking class

Created on 20 Mar 2013

Author: Eleftherios Avramidis

Classes

[hide private]

Ranking
Class that wraps the functionality of a ranking list.

Functions

[hide private]

[int, ...]

indexes(ranking_list, neededrank)
Returns the indexes of the particular ranks in the list

source code

tuple(float, float)

_handle_tie(ranking_list, original_rank, modified_rank, ties_handling)
Modifies the values of the tied items as specified by the parameters

source code

[float, ...]

normalize(ranking_list, **kwargs)
Convert a messy ranking like [1,3,5,4] to [1,2,4,3]

source code

[float, ...]

invert(ranking_list, **kwargs)
Inverts a ranking list so that the best item becomes worse and vice versa

source code

Variables

[hide private]

__package__ = None
hash(x)

Function Details

[hide private]

indexes(ranking_list, neededrank)

source code

Returns the indexes of the particular ranks in the list

Parameters:

ranking_list (list) - the list of ranks that will be searched
rank (float) - a rank value

Returns: [int, ...]

the indexes where the given rank appears

_handle_tie(ranking_list, original_rank, modified_rank, ties_handling)

source code

Modifies the values of the tied items as specified by the parameters

Parameters:

ranking_list (list) - the list of ranks
original_rank (float) - the original rank value
modified_rank (float) - the new normalized rank value that would have been assigned if there was no tie
ties_handling - A string defining the mode of handling ties. For the description see function normalized()

Returns: tuple(float, float)

the new value of the given rank after considering its ties and the value of the rank that the normalization iteration should continue with

normalize(ranking_list, **kwargs)

source code

Convert a messy ranking like [1,3,5,4] to [1,2,4,3]

Parameters:

ranking_list (list) - the list of ranks that will be normalized
ties - Select how to handle ties. Accepted values are:
- 'minimize', which reserves only one rank position for all tied items of the same rank
- 'floor', which reserves all rank positions for all tied items of the same rank, but sets their value to the minimum tied rank position
- 'ceiling', which reserves all rank positions for all tied items of the same rank, but sets their value to the maximum tied rank position
- 'middle', which reserves all rank positions for all tied items of the same rank, but sets their value to the middle of the tied rank positions
inflate_ties (string)

Returns: [float, ...]

a new normalized list of ranks

invert(ranking_list, **kwargs)

source code

Inverts a ranking list so that the best item becomes worse and vice versa

Parameters:

ranking_list ([float, ...]) - the list whose ranks are to be inverted

Returns: [float, ...]

the inverted rank list