A New Estimation Model for Small Organic
Software Project
Wan-Jiang Han, Tian-Bo Lu, and Xiao-Yan Zhang
School Of Software Engineering, Beijing University of Posts and Telecommunication, Beijing, China
Email: {hanwanjiang, lutb, xiaoyan}@bupt.edu.cn
Li-Xin Jiang
Department of Emergency Response, China Earthquake Networks Center, Beijing, China
Email: jlx@seis.ac.cn
Abstract—It is very hard to estimate software development
effort accurately. So far, no model has proved to be
successful at effectively and consistently estimating software
development effort or cost. So it is useful to research a
particular model for a particular type of project. A new
model for small organic project is proposed for software
effort estimation. This model is based on actual project data
and well-established theories, using Gauss-Newton model to
calibrate the parameters of the COCOMO model, using
Fuzzy logic models to maintaining the merits of the
COCOMO model. In particular, this model has been
successfully used in some small project, and has
demonstrated great potential to predict software cost more
accurately.
Index Terms—software cost estimation, software effort
estimation, small project, organic project, Fuzzy,
Constructive Cost Mode
I. INTRODUCTION
As there are a great variety of software development
project in many areas, software estimation is becoming
more and more important in effective software project
management, especially in cost estimation. Accurate
software estimation can provide powerful assistance
when software management decisions are being made; for
instance, accurate cost estimation can help an
organization to better analyze the feasibility of a project
and to effectively manage the software development
process, therefore, greatly reducing the risk.
Lots of attempts [1], [2], [3], [4], [5], [6] have been
made to solve the problem in the last few decades, no
approach has proven to be successful in effectively and
consistently predicting software effort.
So it is useful to research a model for particular type of
project. This paper offers a new model to estimate the
software effort for small organic project based on the data
of actual projects and it is the improvement of COCOMO.
We have taken into consideration the features of the
effort estimation problem and some techniques and have
proposed a new model. We imply Gauss-Newton model
and Fuzzy model to the COCOMO, and have validated
our approach with later project data.
Gauss-Newton model are used in our model to
automatically calibrate the parameters of the COCOMO
model.
II. B
ACKGROUND
Our
new model is based on the standard COCOMO
model, the gauss–newton algorithm and the fuzzy logic
model, we briefly review these techniques.
A. COCOMO Model
The COCOMO model originally published by Boehm
is one of most popular parametric cost estimation models
of the1980s[1] , [11]. At present, the model is still the
most important in the software field. In the middle of
1990’s, Boehm proposed COCOMO II[1], [2] based on
COCOMO81 . Nowadays, it is considered as one of the
most extensively used and approved software estimating
model in academia and industrial area. The basic
principle of COCOMO model is to express effort with
software size and a series of cost factor, as the following
equation:
() ()
B
PM A Size EM
∑
=× ×
∑
∏
B. Gauss–Newton Algorithm
Gauss–Newton algorithm is a method used to solve
non-linear least squares problems. The method is named
after the mathematicians Carl Friedrich Gauss and Isaac
Newton [7], [8].
Non-linear least squares problems arise for instance in
non-linear regression, where parameters in a model are
sought such that the model is in good agreement with
available observations.
Given m functions r = (r
1
,…,r
m
) of n variables β=
(β
1
,…,β
n
), with m ≥ n, the Gauss–Newton algorithm finds
the minimum of the sum of squares, as in (1).
∑
=
=
m
i
i
rS
1
2
)()(
ββ
(1)
Manuscript received October 2, 2012; revised March 7, 2013.
doi:10.4304/jsw.8.9.2218-2222