,

is was ࣢଴: ܨൌ0,

mary(model)$fstatistic

s[1],fs[2],fs[3],lower.tail=FALSE)

he multivariate OLR model constructed for the olive oil content data employing

dent variables. The open circles or ‘In bands’ stand for the observations within

nce bands (the 90% confidence level) while the crosses or ‘Out bands’ stand for

tions beyond the confidence bands (the 90% confidence level).

generalised additive model algorithm

iological data can show simple relationships between variables.

to model the nonlinear relationship between variables, many

or nonlinear regression algorithms can be considered as the

es. Among them, the generalised additive model algorithm

s a good candidate [Hastie and Tibshirani, 1990]. A GAM model

r function of spline functions. The nonlinearity is embedded in

pline function. GAM is defined as below, where ݏሺݔሻ is a spline

of the independent variable x

ݕൌߚ଴൅ߚଵݏሺݔሻ൅ߝ



ne function is a piece-wise approximate of an unknown function

mated. Suppose the data interval is between ݔ௦௧௔௥௧ and ݔ௘௡ௗ and