Pakistan Journal of Nutrition 12 (9): 865-869, 2013
ISSN 1680-5194
© Asian Network for Scientific Information, 2013
Corresponding Author: Muhammad Aman Ullah, Department of Statistics, Bahauddin Zakariya University, Multan 60800, Pakistan
865
Non-Linear Regression Models to Predict the Lamb and Sheep Weight Growth
Muhammad Aman Ullah, Muhammad Amin and Muhammad Ansar Abbas
Department of Statistics, Bahauddin Zakariya University, Multan 60800, Pakistan
Abstract: The study of living organism growth is the hot issue now a days in biological sciences. In this
research, we explore the descriptive analysis of Lamb and Sheep weight growth in Pakistan. Also the non-
linear regression models are identified on the basis of model selection criteria’s for lamb and sheep weight
growth prediction. From the analysis, we have found that for commercial sheep, spring born lambs and twin
born lambs, the best model is the Von Bertalanffy model. While for predicting the weight of male commercial
sheep, autumn born and single born lambs, the best model is the Brody model.
Key words: Growth curve, sheep weight, lamb weight, non-linear regression modeling
INTRODUCTION
Growth of living cells, tissues, organs and organisms is
a biological phenomenon and can be explained in terms
of mathematical terms. Increase in number of cells and
increase in size of cells results in overall increase in
mass of living tissue. This increase is not always steady
and linear but has some non-linear fashion. Therefore,
it may be divided into phases to explain it.
Many researcher defined growth in their own way.
Growth, one of the most essential traits for animals, is
defined as an increase in tissues and organs of the
animals per unit time and effected by genetic and
environmental factors (Tariq et al., 2011). The growth
that has sigmoid form is explained reliably by non-linear
growth models such as Gompertz, logistic, Richards,
Weibull, Monomolecular, Brody and von Bertalanffy.
Information about parameters of these non linear
models enables researchers to obtain beneficial clues
for selection studies.
Growth is a trait of interest in the domestic complex
subject which has been studied through many different
approaches. A widely used approach is to fit growth data
with mathematical equations or growth curve equations.
Those functions are based on deterministic differential
equations that seek a biological interpretation. Even
though growth is a variable among individuals, it follows
a well defined course in populations of animals with
age. Generally; growth follows a sigmoid or s-shaped
curve through the growth rate which varies with age. The
rate slowly declines to zero reaching a plateau when the
animal achieves mature weight (Arango and van Vleck,
2002).
Different models have been developed by researchers
for studying growth and other such attributes. For
instance Benjamin Gompertz (1825) developed
Gompertz model to calculate mortality rates. Today this
model is most frequently applied to study biological
growth. A gompertz curve or gompertz function is a
sigmoid function. It is a type of mathematical model for
a time series, where growth is slowest at the start and
of a time period.
Other important function is the Logistic curve. It was
developed by Verhulst (1838) as a model for population
growth. In this function inflection point is independent on
measurement. It often applied for sigmoid growth where
the inflection is located at approximately half of the
ultimate value and is closely related to the Hubert Curve.
In statistics, logistic regression (sometimes called the
logistic model or logit model) is used for prediction of
the probability of occurrence of an event by fitting data to
a logistic function. It is a generalized linear model used
for binomial regression.
Bertalanffy model was developed by Bertalanffy (1957)
as a model for body weight growth. The point of inflection
is fixed at 8/27 or 29.63% of the maximum value. It is
suitable for sigmoid growth with inflection points around
30% of the ultimate.
In Brody curve the inflection points occurs between the
two curves. It was derived by S. Brody as models for
Piecewise growth process of exponential type. The
increasing function is only valid in temporarily limited
intervals and can be extrapolated. The decreasing
function can be extrapolated and is suitable for
monotonous decrease.
Wildeus (1997) conducted the study to explore the
growth performance of different breeds of sheep in US.
Lancelot et al. (2000) explored the different factors which
are affecting the growth sheep. They use multi-level
modeling and have found that Age, litter-size, age xlitter-
size, litter-sizextreatment and age xlitter-sizex treatment
are the significant factors for growth of sheep. Lamb et
al. (2008) fitted logistic, Gompertz, Richards and
exponential models and linear regression models to
describe the growth of two breeds of lambs from birth to
slaughter. On the basis of model selection criteria’s R
2