es.uam.eps.ir.ranksys.metrics.basic

Class AverageRecommendationMetric<U,I>

java.lang.Object

es.uam.eps.ir.ranksys.metrics.AbstractSystemMetric<U,I>

es.uam.eps.ir.ranksys.metrics.basic.AverageRecommendationMetric<U,I>

Type Parameters:

U - type of the users

I - type of the items

All Implemented Interfaces:

SystemMetric<U,I>

public class AverageRecommendationMetric<U,I>
extends AbstractSystemMetric<U,I>

Average of a recommendation metric: system metric based on the arithmetic mean of a recommendation metric for a set of users' recommendations.

Author:

Saúl Vargas (saul.vargas@uam.es), Pablo Castells (pablo.castells@uam.es)

Constructor Summary

Constructors

Constructor and Description
AverageRecommendationMetric(RecommendationMetric<U,I> metric, boolean ignoreNaN) Constructor in which the average is calculated for all the recommendations added during the calculation.
AverageRecommendationMetric(RecommendationMetric<U,I> metric, int numUsers) Constructor in which the number of users of the recommendation metric to be averaged is specified.

Method Summary

Modifier and Type	Method and Description
void	add(Recommendation<U,I> recommendation) Add a recommendation to the metric.
double	addAndEvaluate(Recommendation<U,I> recommendation) Adds the recommendation metric to the average and returns the user value.
void	combine(SystemMetric<U,I> other) Combines the recommendations added to other system metric to this one.
double	evaluate() Evaluates the metric for the recommendations added so far.
void	reset() Resets the metric by discarding the recommendations previously added.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AverageRecommendationMetric

public AverageRecommendationMetric(RecommendationMetric<U,I> metric,
                                   int numUsers)

Constructor in which the number of users of the recommendation metric to be averaged is specified. Recommendations returning NaN or missing recommendations are treated as zeros to the average.

Parameters:

metric - recommendation metric to be averaged

numUsers - number of expected users' recommendations

AverageRecommendationMetric

public AverageRecommendationMetric(RecommendationMetric<U,I> metric,
                                   boolean ignoreNaN)

Constructor in which the average is calculated for all the recommendations added during the calculation.

Parameters:

metric - recommendation metric to be averaged

ignoreNaN - ignore NaNs from the calculation of the average?

Method Detail

addAndEvaluate

public double addAndEvaluate(Recommendation<U,I> recommendation)

Adds the recommendation metric to the average and returns the user value.

Parameters:

recommendation - recommendation to be added

Returns:

results of the recommender metric

add

public void add(Recommendation<U,I> recommendation)

Description copied from interface: SystemMetric

Add a recommendation to the metric.

Parameters:

recommendation - recommendation to be added to the computation.

combine

public void combine(SystemMetric<U,I> other)

Description copied from interface: SystemMetric

Combines the recommendations added to other system metric to this one.

Parameters:

other - other system metric, should be of the same or compatible class

evaluate

public double evaluate()

Description copied from interface: SystemMetric

Evaluates the metric for the recommendations added so far.

Returns:

result of the metric for the recommendations previously added

reset

public void reset()

Description copied from interface: SystemMetric

Resets the metric by discarding the recommendations previously added.