Genetic {GACFF}

The genetic algorithm for finding similarities between users.

Description

Finding users' similarity by continuous genetic algorithm directly.

Usage

Genetic(ratings, active_user, near_user, Threshold_KNN, max_scour, min_scour,
         PopSize=100, MaxIteration=50, CrossPercent=70, MutatPercent=20)

Arguments

ratings

A rating matrix whose rows are items and columns are users.

active_user

The id of an active user as an integer greater than zero (for example active_user<-6).

near_user

The number of neighbor users that obtained from "NewKNN" for the active user.

Threshold_KNN

Maximum number of neighbors.

max_scour

The maximum range of ratings.

min_scour

The minimum range of ratings.

PopSize

Population size (Number of chromosomes) in Genetic algorithm.

MaxIteration

Number of iterations in Genetic algorithm.

CrossPercent

Percentage of the Genetic algorithm population that participates in the Single-point crossover operator to generate new offspring.

MutatPercent

Percentage of the Genetic algorithm population that participates in the mutation.

Details

The fitness function of the genetic algorithm determines the optimality of the neighbor's similarity to the active user. The fitness function is considered the MAE of the RS. The MAE is obtained by comparing the real ratings of users with the predicted ratings that are calculated according to the similarity obtained by the Genetic algorithm.

The steps of the Genetic algorithm are:

Selection. Selection is based on elitism. Using this operator, the best member of each population survives and will be present in the next population. In other words, the member with the highest match will automatically be transferred to the new population (elitist selection=10\% of the best individuals from each generation). The application of elitism in the genetic algorithm usually improves its efficiency.

Crossover. Single-point crossover technique is used. The crossover operator is used to produce children. A weight coefficient (crossover probability) of between 0 and 1 (0.8) is considered equal to the length of the parent, and by using follow formulas, two new chromosomes or two children are created.

y1=\alpha * x1 + (1-\alpha) * x2

y2=\alpha * x2 + (1-\alpha) * x1

x1 and x2 are decimal values that represent the parent chromosome. \alpha is the weighting factor and, y1 and y2 are the children's chromosomes resulting from the parent compound.

Mutation. Single-point mutation technique is used to introduce diversity. The mutation probability is 0.02. The Gaussian mutation operator is implemented. First, a chromosome is randomly selected from the population, and then one or more of its components is changed according to the Gaussian function using follow formula.

y1= x1 + r1 * N(0,1)

x1 is the similarity value that represents the parent chromosome, r1 is a random number in the range of 0 and 1 (0.02). N(0,1) is a random number distributed by using the Gaussian distribution.

The genetic algorithm stops when an individual in the population has a fitness value less than a constant value (for example 0.5).

Value

An object of class "Genetic", a list with components:

call

The call used.

sim_GA

Similarity obtained from the Genetic algorithm as much as the number of neighbors.

pre_GA

Predicted active user ratings for all items.

item_GA

A set of best-predicted items for the active user.

save_MAE_GA

A set of MAEs obtained from each iteration of the Genetic algorithm.

time_Genetic

The elapsed time of Genetic algorithm.

Author(s)

Farimah Houshmand Nanehkaran

Maintainer: Farimah Houshmand Nanehkaran <[e-mail]>

References

Bobadilla, J., Ortega, F., Hernando, A. and Alcala, J. (2011). Improving collaborative filtering recommender system results and performance using genetic algorithms. Knowledge-based systems, vol. 24, no. 8, pp. 1310-1316.

Examples

ratings <- matrix(c(  2,    5,  NaN,  NaN,  NaN,    4,
                    NaN,  NaN,  NaN,    1,  NaN,    5,
                    NaN,    4,    5,  NaN,    4,  NaN,
                      4,  NaN,  NaN,    5,  NaN,  NaN,
                      5,  NaN,    2,  NaN,  NaN,  NaN,
                    NaN,    1,  NaN,    4,    2,  NaN),nrow=6,byrow=TRUE)

NewKNN.out  <- NewKNN (ratings, active_user=6, Threshold_KNN=4,
                      max_scour=5, min_scour=1)
                      
Genetic.out <- Genetic (ratings, active_user=6, 
                        near_user=NewKNN.out$near_user, 
                        Threshold_KNN=4, max_scour=5, min_scour=1,
                        PopSize=100, MaxIteration=50, CrossPercent=70, 
                        MutatPercent=20)