dgamma(x,shape,rate,···)

two distributional statistics can be estimated using the density

nction named as fitdistr in the MASS library before calling

mma function. The function is a generic one to estimate the

onal statistics for a variety of density functions. Given a vector

by x, the syntax of this function is shown below, where densfun

ameter to specify a density function such as Gaussian or Gamma

fitdistr(x,densfun,···)

tumour area feature of the breast cancer diagnosis data set was

to follow a Gamma distribution, the following code was used to

two distributional statistics (the shape and the rate parameters)

eature, where malignant was generated using the same way as

r the tumour radius

del=fitdistr(malignant,densfun=‘gamma’)

tatistics of this function include the shape and the rate parameters

mma density function. Next, the dgamma function was called to

a Gamma density function for this breast tumour area feature,

model$estimate[1]

odel$estimate[2]

ma(sort(malignant),shape=shape,rate=rate)

e 2.9 shows the resulting densities of the benign tumours and the

t tumours using this method for the tumour area feature of the

ncer data. It can be seen from both plots that each estimated

sing the parametric approach fits the raw density estimated using

gram approach very well. This shows that if the density format is

n advance, the parametric approach is a good alternative to

a density function for a data set. The major advantage of the