y

the cow milk concentration percentage with a very high accuracy.

words, the spectra profile expressed by the short-term Fourier

m variables can be a very good predictor of the milk concentration

ges in this data set.

The Jackknife test of the neural network regression models constructed for the

for the relationship between the spectra pattern represented by the short-term

nsformed data and the cow milk percentage.

yse the bacterial and macrophage data

was used for discriminating between the bacterial and the

age based on the extracted chemicals from the Raman spectra data

al., 2012]. The macrophage was J774.1 and the bacterial was

la enterica serovar Typhimurium SL1344. There are 11

s of macrophage spectra and nine replicates of bacterial spectra

at 30, 90, 150 and 360 minutes. BWH has discovered important

hich match the expectation.

e 5.23 shows the spline estimation of the baselines for 11

age spectra. The advantage of the spline approach is its speed.

, the estimated baselines may have an accuracy problem.

e 5.24 shows the extracted peaks for the first three of 11 replicates

phage spectra based on the spline models. It can be seen that the

d peak spectra contain too many artifacts. In addition, the peak