Use and abuse of chemometrics
Erno¨ Pretsch1, Charles L. Wilkins2

Trends in Analytical Chemistry, Vol. 25, No. 11, 2006
Editorial

Chemometrics, a term originally coined by Svante Wold in
1971 in the title of a research proposal [1], is ‘‘a chemical
discipline that usesmathematics, statistics, and formal logic
(a) to design or select optimal experimental procedures;
(b) to provide maximum relevant chemical information by
analyzing chemical data; and, (c) to obtain knowledge
about chemical systems’’. According to this broad definition,
a leading textbook on the topic [2] indicates that there
is hardly any quantitative analytical method that does not
make use of chemometrics. Even if one confines the scope to
multivariate data analysis and softmodeling, chemometric
techniques are increasingly being applied in many fields of
analytical chemistry. Often, ready-to-use toolboxes are
available to develop models using various techniques,
including principal component analysis (PCA) and related
methods, neuronal networks, experimental design, and
genetic algorithms. There are analytical techniques, near
infrared spectroscopy being a prototype, that would be
unimaginable without chemometrics. Thus, one could
celebrate a triumph of chemometrics.

So what is wrong? There is a two-fold problem with
today's use of chemometrics. First, many sophisticated
techniques are applied by uninitiated scientists who do
not have a fundamental knowledge of the capabilities
and the limits of the chosen method. It is then sometimes
like a contagious illness â€" the more people who use an
inappropriate method or model, the more probable it is
that others will also apply it. As is shown in one of the
articles in this Special Issue, literally thousands of
incorrect data treatments are published, just because
others do it the same way. One goal of this Special Issue
is to highlight some typical pitfalls.

The second problem with chemometrics is that many
novel efficient methods are not applied in practice just
because they are not implemented in the software of
commercial analytical instruments.

Of course, a small volume such as this cannot treat all
relevant problems comprehensively, so a somewhat
arbitrary selection was made in order to illustrate key
problems. Nevertheless, several actual developments,
including image analysis and data treatment in the
‘‘omics’’ fields, are critically reviewed. In addition, there
is critical discussion of established, but often incorrectly
used approaches, including simple linear models or the
use of the Horwitz equation to estimate the precision of
data on the basis of the total concentration.
There are hundreds of other applications or proposed
applications, especially in such fields as biomarker discovery,
with a high false-positive identification rate, as
addressed in the present issue.

There is no doubt that unwarranted reliance on poorly
understood chemometric methods is responsible for some
of these problems. It is our hope that the current collection
of articles can make clear what difficulties and
incorrect applications exist and how to avoid them. It is
our strong belief that chemometric methods cannot be
dismissed from the science of measurement, and their
importance is steadily increasing. Thus, in spite of the
somewhat provocative title, the goal of this Special Issue
is to provide information that will help to avoid misuses
of chemometrics.


References
[1] R.G. Brereton, Chemometrics, Ellis Horwood, Chichester, West
Sussex, UK, 1990.

[2] D.L. Massart, B.G.M. Vandeginste, L.M.C. Buydens, S.D. Jong,
P.J. Lewi, J. Smeyers-Verbeke (Editors), Handbook of Chemometrics
& Qualimetrics, Elsevier, Amsterdam, The Netherlands, 1997.

lho kok di math ini??? hihi... ngantuk yahh

-----------------

dah dipindah

hehe ... bukaaan, baru baca paragraph pertama waktu posting  ... emang bingung sih, di chemistry atau math? tapi ini problem math lho, chemometrics ini problem statistics, cuma aplikasinya aja yg bisa di kimia/farmasi/biologi

wah, saya yang ngantuk berarti.... dah dibalkin lagi nih. Coba bentar baca lagi.

Maksudnya metode chemometric yang sering menyebabkan false positif yah ? Ya udah, ngikut yang posting aja, biar di sini berdasarkan ini aja :

Kutip
First, many sophisticated techniques are applied by uninitiated scientists who do not have a fundamental knowledge of the capabilities and the limits of the chosen method.

lho dibalikin lagi? ... sebenarnya aku juga bingung   heheheh ... bisa di dua2nya kayaknya ... dan emang bener lagi ngantuk sih, kopinya nggak mempan tadi  

terserah admin lah mo ditaruh di mana, pasrah......

btw, sorry post pertama cuma editorial aja ... baru nemu review yg kayaknya lumayan, kalo2 mo baca lebih banyak ... tp 976KB wuaaa, ditolak dah ama FS      

ntar dulu deh, cari cara bagi pdf jd 3 file ... kayanya gak bisa ya kalo cm pake acrobat reader?

KutipMaksudnya metode chemometric yang sering menyebabkan false positif yah ?

iya, tapi bukan salah metodenya sih ... false positif itu krn user-nya, kita2 yg bukan statisticians, asal make aja

Kutip dari: peregrin pada Februari 17, 2007, 02:23:47 AM
lho dibalikin lagi? ... sebenarnya aku juga bingung  heheheh ... bisa di dua2nya kayaknya ... dan emang bener lagi ngantuk sih, kopinya nggak mempan tadi 

terserah admin lah mo ditaruh di mana, pasrah......

btw, sorry post pertama cuma editorial aja ... baru nemu review yg kayaknya lumayan, kalo2 mo baca lebih banyak ... tp 976KB wuaaa, ditolak dah ama FS     

ntar dulu deh, cari cara bagi pdf jd 3 file ... kayanya gak bisa ya kalo cm pake acrobat reader?

Ikut ajalah... sama aja  ... udah dinaikin max nya jadi 2MB 

Kutipudah dinaikin max nya jadi 2MB 

asyiiiik    ... jd nggak usah pusing2 mecah file 

admin memang selalu mendengarkan aspirasi para membernya ... ntar kapan2 kalo mo nyalonin diri jd presiden aku pilih deh, pasti 

btw, aq gak ujian tissue engineering ... kalo yg "omics" itu iya, rada related ama kerjaan yg skr

kita kok jd kaya chat ya ... pake pindah2 thread lg tp nyambung melulu ... yg lain pada ke mana? tidur yaa     

Kutip dari: peregrin pada Februari 17, 2007, 02:37:52 AM
ntar kapan2 kalo mo nyalonin diri jd presiden aku pilih deh, pasti 

abis quote anti pemerintahan n nyela2 pejabat trus calon presiden? Indo banget 

Kutipbtw, aq gak ujian tissue engineering ... kalo yg "omics" itu iya, rada related ama kerjaan yg skr

kita kok jd kaya chat ya ... pake pindah2 thread lg tp nyambung melulu ... yg lain pada ke mana? tidur yaa     

emang kerjaan yang skrg apa?

nah, makin kayak chat kan pertanyaannya 

Kutipabis quote anti pemerintahan n nyela2 pejabat trus calon presiden? Indo banget 

gaya akbar tanjung dong 

Kutipemang kerjaan yang skrg apa?

nah, makin kayak chat kan pertanyaannya   

ya si admin lagi iseng nih  ... kagak ada teman yg bisa digedor2 tengah malam ya 

Wah, asal copy yaaa


Mencontohkan hal yang ga baik tuch