Conclusions and discussion
In this paper, we propose a new decision support system to
solve the problem of hen allocation to hen houses with the aim
to minimize the total cost. Clustering of hen houses was first determined
to allocate hens to hen houses effectively. The problem of
hen allocation to hen houses has many similarities to the Growing
Neural Gas algorithm (GNG), in which groups of similar hen houses
will be clustered together. However, the traditional GNG often
solves the clustering problem by considering distance only. Therefore
the hybrid Growing Neural Gas (hGNG) was proposed by considering
both the distance from the centroids of the clusters to the
hen houses and the weights of hen house sizes. In the second
phase, allocating and determining routes to allocate hens to the
hen houses using the nearest neighbor approach were carried out
in order to minimize the total distance. The performance of the
algorithm was measured using the relative improvement (RI),
which compared the total cost consisting of three components
being the cost of farm utilization, hen transportation cost, and loss
from mixing hens at different ages. The results obtained from this
study showed that the hGNG algorithm provided better total cost
values than the firm’s current practice by 7.92–20.83% (with an
average of 14.56%), and 5.90–17.91% (with an average of 11.14%)
better than the traditional GNG algorithm.
The statistical results obtained from ANOVA and Duncan’s test
showed that the RI increased as the number of hen houses
increased. This implies that a higher number of hen houses would
most probably yield high flexibility for the hGNG algorithm to allocate
hens to hen houses effectively. Therefore, a long-term planning
horizon can be beneficial for reducing the number of hen
houses used and especially the loss from mixing hens will be
decreased significantly. Additionally, paired sample tests based
on the values of RI were performed to compare the total cost of
the proposed hGNG and the traditional GNG algorithm, which
showed that the proposed hGNG was significantly better than
the traditional GNG. Hence, the hGNG algorithm was used to compare
the total cost with that of the current practice for the algorithm’s
effectiveness. The result on paired sample tests indicated
that the hGNG algorithm can be used efficiently to solve the problem
of hen allocation to hen houses. Therefore, the proposed
method developed in this paper will help the egg industry reduce
the total cost resulting in economically sustainable production.
The proposed hGNG method is not only useful for reducing the
total cost when compared to the current practice, but also for efficient
management of a poultry production system. Furthermore,
the method of this research should prove beneficial to other similar
agro-food sectors in Thailand and around the world. Therefore,
there is still much greater opportunity to extend our work in many
areas. Our planned future work will be to improve the routes from
pullet houses to hen houses in order to minimize the transportation
cost. Another valuable avenue for future research is to consider
some other parameters of the problem as fuzzy variables,
such as egg demand. We believe that this can add to the ability
of our method to model real world problems and will be a valuable
extension. Additionally, although the hGNG algorithm has shown
an outstanding ability to solve the problem at hand, there is a possibility
to use hybrid methods or other meta-heuristics to solve the
same problem or to conduct an empirical study to compare the
strength of various approaches in solving other problems of this
nature.