Chemometrics in Australia

What is Chemometrics? | Australian Research | Web Links | Journals | Tutorials | Additional Resources | Software | Experimental Design: References/Downloads


What is Chemometrics?

"Symbol of the International Chemometrics Society"

ICS Symbol

According to the International Chemometrics Society (ICS), "Chemometrics is the science of relating measurements made on a chemical system or process to the state of the system via application of mathematical or statistical methods."

Back to Top


Chemometric Research in Australia:

Here is a list of active groups and individuals in chemometrics research in Australia. As it is a new field the list is short and probably incomplete. Suggestions of additions for the list would be most welcome.

University of New South Wales
Chemometrics and Metrology Group
Prof. Brynn Hibbert

Electroanalytical chemistry, Sensors, Electronic nose technology, Self assembled monolayers and biosensors, Electrodeposited fractals and mechanistic electrochemistry, Chemometrics, Multivariate analysis of data, Artificial intelligence applied to chemistry, Non-linear dynamics and chaos.

University of Newcastle
Dr. Marcel Maeder

Non-linear least-squares fitting of complex, multivariate chemical measurements and application in equilibrium and kinetic studies and hyphenated chromatography. Analysis of applied andindustrial data using PCR, PLS and related techniques.

RMIT University
Chemometrics and Instrument Optimization Group
Prof. Mike Adams
Dr. Jeff Hughes
Prof. Philip Marriott

Application of experimental design techniques to the optimisation of chromatographic techniques for the analysis of complex mixtures.

Use of pattern recognition methods to analyse complex environmental data sets. Applications of chemometric methods to qualitative and quantitative spectroscopic analysis of polymers and environmental samples.

Back to Top


Collections of General Chemometrics Links:

There are a number of other sites with good collections of links:

Back to Top


Journals:

Back to Top


Tutorials:

The Web is an increasingly expanding source of tutorial material on many aspects of Chemometrics.

One of the best collections of informative articles has been published by Richard Brereton in The Alchemist, part of the ChemWeb site. You need to register to access this site but registration is free. A complete list of articles (1999-2003) can be found at the Bristol Chemometrics site.

Johan Trygg's Homepage of Chemometrics has a good list of tutorials also. The site also has access to discussion forums, a Chemometrician's address book (with more than 200 entries) and a searchable database of more than 500 publications of Svante Wold's research group.

A great resource is the Electronic Statistical Textbook published by Statsoft. This text has chapters on all aspects of Statistics andincludes chapters on Experimental Design, Cluster Analysis, Factor Analysis and Neural Networks. The complete set of html documents can be downloaded for easy local access.

Another online textbook is the Engineering Statistics Handbook, a very comprehensive resource in html format. The individual chapters can also be downloaded as pdf files. The Dataplot software can also be downloaded from this site.

A short primer Chemometrics for spectoscopists - this is a good introductory tutorial by Professor Stephen Brown in the application of modelling techniques in spectroscopy. This tutorial is part of the Chemometrics section of the Spectroscopy Now site, maintained by the publisher Wiley.

Some other useful tutorials include:

Back to Top


Additional Resources:

Back to Top


Software:

Minitab is an excellent general-purpose Statistics program, widely used in Universities. It is particularly user-friendly and has good facilities for Experimental Design and Principal Component/Factor Analysis / Partial least squares. A demo version (30 days) and student statistics tutorials can be downloaded from the site.

Statistica is another general statistical program. Add-on modules can be obtained for DOE and Neural Network analysis. A scaled down version of the program, along with electronic guide and guided tour movie, can be downloaded. The electronic textbook, described above, is also available from this MATLAB, a popular and powerful tool for exploring data and creating algorithms. Extensive online toolboxes are available for applying chemometric techniques.

Strategy is a software program for Design and Analysis of Experiments. It has a user-friendly interface and a comprehensive range of analyses. A 30-day trial version can be downloaded and is a good way to get a feel for how DOE operates.

Camo products include The Unscrambler and the Product Optimizer, multivariate data analysis and experimental design software. The Unscrambler offers powerful methods like PCA, PLS, 3-way PLS and data pre-treatment including SNV and ESMC. The Product Optimizer is an intuitive optimisation module for product formulations and process optimisation and is based on modules calculated using the Unscrambler.

Hearne Scientific Software have demo versions of Systat, Axum, @Risk, SigmaPlot and MathCad available for download. A demo version highly recommended is Neuralyst for Excel, an excellent introduction to Neural Networks. The demo is not time limited and the program is quite easy to use. A good tutorial comes with the demo.

Pirouette software have a downlaodable demo of their good, general-purpose chemometric software.

Design-Expert DOE software A 30-day demo of this excellent DOE softare can be downloaded

SCILAB is free software which provides an open computing environment for scientific applications. A multifractal-wavelet analysis toolbox is available

Mathematica and MathCad are general purpose software platforms providing symbolic and numeric computational engines and graphical visualization. Packages are available for both programs to implement chemometric routines.

Extract is designed for visualization of the structure and quantitative relationships in data and modelling of these relationships with multivariate statistical methods.

Sirius and Xtricator were developed by the Pattern Recognition System (PRS) company. Sirius is a tool for multivariate analysis and experimental design. Xtricator was developed for analysis of data from hyphenated analytical systems such as GC-MS.

The Chemometrics Add-In for Excel can be downloaded from the Bristol Chemometrics site. The package features a user-friendly interface for PCA,PCR,PLS and MLR and is highly recommended.

ParLeS Shareware for Chemometric Analysis of Spectroscopic Data. The software can perform PCA,PLS modelling and prediction with n-cross validation, bootstrap aggregation PLS and calculates various assessment statistics. Numerous preprocessing and pre-treatment algorithms are available.

Back to Top

Spreadsheet Applications

The aim here is to provide a collection of spreadsheets applying some aspect of chemometrics, which can be downloaded.

1. 'How many pills from a drug seizure should we anlayse? - A Bayesian Approach' This spreadsheet, using a macro developed by Brynn Hibbert, calculates the probability that a siezure of size N contains M positives, given that a sample of size n contains m positives. It also calulates how many pills from the seizure need to be analysed so that, if all the sample are positive, we can conclude all the sieizure are positive.

Bayesian Pills Calculator

2. This spreadsheet calculates errors for the 'inverse calibration' problem. This is the situation, common in analytical chemistry, where you want to predict the error in x from a measured y (e.g. concentration from absorbance), using a calibration graph developed from regression of y on x (e.g measured absorbances for a set of concentrations). The spreadsheet applies this to the errors from the standard additions method and also calculates limits of detection for an analytical method, using the regression line.

Inverse Calibration Spreadsheet

Back to Top


Experimental Design:

"Response Surface Plot"

This is an example of a response surface plot for the optimization of capillary electrophoresis separation of a mixture of ranitidine-related compounds. Ranitidine is a drug used in the treatment of gastric and duodenal ulcers. The two variables being optimized are pH and applied voltage. The response variable used in the optimization is the logarithm of the CEF function, a parameter devised to assess the quality of a chromatogram. This function takes into account peak separation and total elution time.

References:

The following textbooks provide a good general coverage of this topic:

A First Course in Design and Analysis of Experiments, G. W. Oehlert (Freeman, New York, 2000)

Chemometrics: Experimental Design, E. Morgan (Wiley, London, 1991)

For a more theoretical treatment the following references are useful:

Response Surfaces: Design and Analysis, A.I. Khuri and J.A. Cornell (Dekker, New York, 1996)

Empirical Model-Building and Response Surfaces, G.E.P. Box and N.R. Draper (Wiley, New York, 1987)

The following review articles also provide good introductory summaries:


Attaining Optimal Conditions, H.V. Ryswyck and G.R. van Hecke, J. Chem. Ed., 68, 878 (1991)

Introduction to Experimental Design for Chemists, R.S. Strange, J. Chem. Ed., 67, 113 (1990)

Central Composite Design Experiments, J.A. Palasota, S.N. Deming, J. Chem. Ed., 69, 560 (1991)

Experimental Design and Optimization, T. Lundstedt, E. Seifert, L. Abramo, B. Thelin, A. Nystrom, J. Petterson and R. Bergman, Chemometrics and Intelligent Laboratory Systems, 42, 3 (1998)

Experimental Design 1. Screening and 2. Optimization, P. W. Araujo and R. G. Brereton, Trends in Analytical Chemistry, 15, 26-31 and 63 - 70 (1996)

Introduction to Statistical Experimental Design (Aug 2002, Johan Trygg)

The online texts mentioned above also have good chapters on experimental design (the DOE chapter in the Electronic Stats textbook and chapter 5 in the Engineering Stats Handbook)

Downloads for Experimental Design:

There are some very good software packages on the market for experimental design and analysis but they can be quite expensive and can be a daunting obstacle for anyone wanting to start in DOE. However designs such as factorial and central composite designs can be constructed and analysed in spreadsheets such as Excel. A tutorial in an Excel spreadsheet that demonstrates use of full and fractional factorial designs is available for download. Also available are some lecture notes from a chemometrics course on data treatment and experimental design.

Excel spreadsheet illustrating factorial designs

PDF Files :

Chemometrics Unit 1: Measurement and Assessment of Variability
Chemometrics Unit 2: Regression Analysis
Chemometrics Unit 3: Experimental Design - Factorial Designs
Chemometrics Unit 4: Experimental Design - Response Surface Methodology

Back to Top

Send any questions/comments/corrections on these pages tojeff.hughes@rmit.edu.au