You are on page 1of 4

20IO International Coriference on Educational and Information Technology (ICEIT 2010)

Research on Human Resource Allocation Optimization Based on Genetic Algorithm


From the Perspective of Two-way Choice Model

Yi Zhang
School of Management;
Inner Mongolia University of Technology;
Huhhot, 010051, China
zhangyinmg@163.com

Abstract-This rational and efficient allocation of human


resources is a prerequisite to improve overall efficiency of the II. TwO -WA Y CHOICE MODEL OF HUMAN RESOURCES
organization. This article analysed and evaluated the related OPTIMIZATION BASE D ON GENETIC ALGORITHM
elements of organization, and then it proposed a two-way
choice model of optimal allocation of human resources form A. Problem Analysis
the maneuverability. We used the genetic algorithm to solve
The corporate personnel structure is composed of the
the model and desgined detail process of the algorithm. That
staffs of different skills, different qualities in different
method provided a new specific quantitative management tool
positions within the enterprise. This model' s idea is that we
to optimal allocation of human resources. Finally, a numerical
example is calculated to proved the effectiveness of the methos.
will build a two-way choice model which is based on staffs
and enterprise mutual participation to improve the
Keywords- genetic algorithms;human resources; model organization's overall performance through the staff
quantitative evaluation and post synthesis analysis; we called
this model two-way choice model. This model could make
1. INTRODUCTION
the staff and the various positions find the best counterparts
Human resources do not only relate to the enterprise's and then to achieve optimization of personnel structure [4).
survival and development, but also relate to the improvement Two-way choice is that enterprise chooses employees,
of enterprise's core competitiveness and the viability. The while employees either choose option positions [5]. In this
essence of human resource allocation is that managers put paper, the human resources allocation is a decision-making
human resources into the various jobs within the process which combines the objectives of enterprise and
organization so that human resources combine with material employees. On the one hand, human resources allocation
resources and corporate organizational goals combine should enable companies to achieve organizational
employees' personal goals [I). Human resources allocation objectives; the other hand, human resources allocation
is not only the end. The ultimate aim of human resources should help employees to achieve personal goals.
allocation is to reach a station which staffs match post very
B. The Mathematical Model of Human Resources Optimal
well, to enhance the organization's overall performance.
Studies have shown that the labor productivity of the best Allocation

employee is three times higher than the worst staff on the Two-way choice mathematical model should consider
same position than. However, in an organization, that a that enterprise chooses staffs and staffs choose posts two
specific post will need to be allocated the person should be positions. This article described the mathematical model of
optimally chose according to the system' s overall, rather human resources optimization described as follows: assign
than choose the best employees to one post [2). This means M staffs to N different type's posts. Each type post
that the allocation of personnel need a good identification should be configured a certain number of employees. And
and selection process to selected out the right staff which is we could solve the biggest revenue optimization solutions on
of corresponding skills, knowledge and experience. At the the conitions of meeting the various constraints. First,
same time that allocation will achieve organizational system enterprises require staffs to meet the a variety of conditions
optimization. However, in the actual operation of human about the post, such as basic quality, work experience,
resources allocation, it often did not reach the desired results education and other qualifications; Second, when staffs
for lack of scientific and feasible method. Through select post they would consider that the post whether meet
discussion and analysis of the optimal allocation of human employees' interests, goals, skill and other requirements. The
resources, this paper developed a two-way selection model mathematical model of human resources allocation is as
of optimal allocation of human resources, and the used follows [6]:
advanced genetic algorithm to solve for the model. The
n M
method achieves to meet the system-optimal needs of
optimal allocation of human resources [3].
max z(x) = LLE; x Pu xqu x Xu (1)
;�1 j�O

978-1-4244-8035-7110/$26.00 © 2010 IEEE VI-380


2010 International Conference on Educational and Information Technology (ICEIT 2010)

q ij : The comprehensive scoring of staffs evaluates post


(2)
whether meet the staffs' requirements.
Decision variables;

{3ir(bijr � di� )
= tt
i r dIr
(3) {I;
X.I . =
if j staff is assigned to post i;

n M
f 0; otherwise.
S.t. GO (X) =
II jXij � M, (4)
i=1 j=O
III. GENETIC ALGORITHM FOR SOLVING
n

Gi(x) = Xij =1,I Vi, (5) A. Genetic Algorithm Description


j=1
Genetic Algorithm (GA) is perhaps the most well­
known of all evolution based search technique. GA is a
search algorithm based on natural selection that transforms a
set of individuals within the population of solutions into a
(6)
new set for the next generation using genetic operators such
as crossover and mutation. A survival of the fittest strategy is
adopted to identifY the best solution and subsequently
s {3I.r(bljr - dOIr )�
genetic operators are used to create new solutions for the
next generation. This process is repeated generation after
0
I dIr 0
(7)
generation until a satisfactory solution is found. Genetic
algorithms have been successfully used to obtain solutions
Xij =0 or 1, V i, j. for many combinatorial optimization problems.
th Genetic algorithm for a combinatorial optimization
Where, Ei IS the essentiality of the i post, problem works as follows: The search space contains all the
·th search nodes for the given problem. GA starts with an initial
( j =1,2, ... ,n); ail is the weight of the 1 post's
population of search nodes from the search space. Each
th
z competency; ( j=1,2, ···,n;/=1,2, ···,M); dij, is the search node has a fitness value assigned to it using the
objective function. New search nodes are generated for next
h th
level of the / post's / competency; (/= 1,2, ... ,M; generation based on the fitness value and applying genetic
operators to the current search nodes. These processes are
l�dijl �9, dijl E N) ; dit 0 is the reference level of repeated for eneration after generation until the algorithm
·th th /- converges.
1 post's / competency; - 1,2, ,M;
1 �duo �9, di/ 0 E N) ; , is the weight of the i
th B. Coding Design

Because each employee correspond one post in the


th . bijr optimal allocation programs of human resources, every
Post's r competency·, (J=1 2 ... n·, r= 1 2 ... M)· is
' " " "
chromosome contains a large amount of information. This
·th
the level of the J post's r competency·, (/_
th paper used the virtual value for encoding; each chromosome
-I 2 ··· M· " "
shows a group of optimization solutions. Chromosome's
1 �bijr �9, bijr E N) ; biro is the reference level of genes sites denote the number of different types post; gene
values represent a virtual value of optimization solutions on
·th th
1 post's r competency; ( /-
- 1,2, ... ,M; the post. Chromosome's encoding is shown in figure I, the

l�bi ro �9, bi roE N).


·th
value of the 1 bits (i
=I, 2, ... ,) is the optimal solution
Parameter: values on the post. The virtual value is determined by the
i index of post, (i=1 2 ... N)· " "
binary system the values which represent the optimal
allocation on the post translation into decimal values; On
j j each gene site's binary-code is composed of number '0' or
index of staffs, ( =1, 2,. ··,M);
n total number of post; 'I " the length of the number is 'n' which is the number of
employees. Binary digit'I' represents assign the staff to the
m total number of staff.
post; specific codes are as follows:
Pij
: The comprehensive scoring of enterprise evaluates
staff whether meet the post requirements.

VI-381
2010 International Coriference on Educational and Information Technology (ICEIT 2010)

For example: The enterprise expected to allocate 10 SteplO: end.


staffs to 3 posts. One of the codes for a chromosome is P:
[010 100 10 10110 10 100000100000 10 100] The
IV. THE CASE

In this paper, we studied the human resource allocation


meaning of this chromosome is that the enterprise allocate optimization. The corporate's manager will assign 12
the employees No. 2, 4, 7, 9 to post 1; and staffs No. 1, 3, 5 employees to four different posts. This paper adopted the
will be assigned to the post 2; No. 6 and 8, will be assigned subjective scoring method to evaluate the quality (requires)
to the post 3. of every posts with figures 1:9 to describing that, where '1'
shows the lowest level requirements, '9' shows the highest
TABLE! . CHROMOSOME'S ENCODl�
Chromosome level of requirements. And then we could obtain the
site evaluation data table. The scoring of enterprise evaluating
2 3 4 5 N Chromosome
value
staffs is shown table 1, and the quality requirements of
various posts are shown in table 2. The scoring of enterprise
evaluating staffs is shown as table3 (omitted).

TABLE!! . THE SCORING OF ENTERPRlSE EVLUA nON STAFFS


C. Genetic Operation
Basic
Options: Commonly used selection operators have the Staff Experience Skills Education
quality
following categories: fitness proportional method, the best
individual preservation law, the expected value method, I 5 4 6 5
ranking selection method, league selection and exclusion law.
In this paper, we adopted the best individual preservation 2 4 4 6 5
strategy and roulette wheel method to generate a sub­
population, in the older generation, individuals whose fitness 3 5 4 7 4
is higher (10%) direct access to the next generation of groups
4 4 5 2 8
without crossover and mutation operations.
Cross: There are lots of methods for cross, such as single­ 5 5 5 7 5
point crossover, uniform crossover, partially matched
crossover, order crossover, cycle crossover. Aming at 6 6 4 8 6
encoding for this article, we used the single-point crossover.
Variation: The standard mutation operators have the 7 6 7 2 4
following three kinds:
1) The basic mutation operator. 8 5 4 6 4
2) Reverse the mutation operator.
9 7 5 5 4
3) Select point transposition operator.
Aming at encoding for this article, we used the basic
10 4 5 6 6
mutation operator.
II 7 5 4 4
D. Genetic Algorithm Steps

Step1: GA parameter initialization, determine the size of 12 5 7 4 8


the initial population POP, crossover probability is P
crossover, mutation probability is P mutation maximum
number of iteration is K. TABLE IlL THE QUALITY REQUIREMENTS OF VARlOUS POSTS
Step2: chromosome random assignment;
Step3: the number of iterations k=l, n=l. Basic
Step4: If n<POP, then n=n+1. Pick out 'n' chromosomes Post Experience Skills Education
quality
from the initial population to decode and calculate the fitness
of all chromosomes F. I 4 4 5 5
Step5: select chromosomes according to fitness by
roulette wheel method. 2 3 4 4 5

Step6: generate the next populations; pick out any two


3 4 5 4 5
chromosomes as the male parent, cross operator in
accordance with cross-operation; 4 3 5 5 3
Step7: pick out any chromosome to implementate
mutation.
Step8: after crossover and mutation operator, re­ In order to prove the efficiency of the model this paper
calculation of fitness value of chromosome. used the matlab 7.0 to make a simulation. Based on a large
Step9: If k<K, so k=k+l, repeat Step5; otherwise number of experimental, the chromosomes' optimum
implementate Step1O. parameters are as follow: W makespan =0.8, W cost=0.2,

VI-382
2010 International Coriference on Educational and Information Technology (ICEIT 2010)

POP=60, P crossover=O.7, P mutation=O.05, K=6000. We REFERENCES


can get a statisfactory result with the algorithm which has [I] Lagesse, B. A, "Game-theoretical Model for Task Assignment in
been validated by many times experiment. After the Project Management," Management of Innovation and Technology,
simulation of GA, we can obtain the optimal solution of this 006 IEEE International Conference on Volume 2, Issue, June
IS:
problem- the optimal chromosome solution 2006 ,pp.678-680

[230411090105561144]' after translation from decimal to [2] J Kennedy, RC Eberhart, "Particle Swarm Optimization, in:Proc.
Of IEEE Int ConY. on Neural Networks, Perth, Australia , 1995,
pp. 1942- 1 948
binary, we can get the solution is J Kennedy, RC Eberhart, "A New Optimizer Using Particle Swarm
[100 10000000110 1000 1000010100 1000 1 a 1100
[ 3]
Theory, in Proc, of the Sixth Int Symp. On Micrs Machine and
Human Science (MHS' 1995), Nagoya, Japan, 1 995, pp. 3 9-43

1000010010000] . So the manager of the enterprise would [ 4] L. Barbulescu, J.P. Watson, L. D. Whitley, A E. Howe, "Scheduling
Space-ground Communications for the Air Force Satellite Control
allocate the staffs No. 1, 4, 12 to the first post, and allocate Network," Journal of Scheduling, VoLl, No.I ,2004,pp.7- 34
the staffs No. 2, 6, 11 to the second post, and allocate the [ 5] WI. Wolfe, SE Sorensen, "Three Scheduling Algorithms Applied to
staffs No. 3, 7, 9, 10 to the third post; and allocate the staffs the Earth Observing Systems Domain," Management Science, VoIA6,
No. 1,2000, pp. 148- 168
No. 5, 8 to the third post.
[ 6] A Globus, J Crawford, J Lohn, A, Pryor, " Scheduling Earth
V. CONCLUSIONS Observing Statellites with Evolutionary Algorithms," International
Conference on Space Mission Challenges for Information
This paper developed a mathematical model of human Technology, Pasadena, CA, JulY,2003, pp.34-56
resources allocation with considering the two-way staffs and
enterprise. This method embodies the double principles of
staffs' competency mathching with post request and staffs'
expectations matching with post, which according with the
actual situation of human resource management. Besides,
application of GA to slove human resource allocation
optimization problems showed sufficiently the advantages of
GA which is of high quality for solving and is of high speed
for convergence. And after an example simulation, the
method has been proved useful and effective.

VI-383

You might also like