This course is designed to get the student quickly "up to speed" with the Discriminant Analysis and Factor-Based techniques used to produce qualitative calibrations from instrumental and other data: Euclidean distance, Mahalanobis distance, K Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA), Piecewise Linear Discriminant Analysis, Artificial Neural Networks (ANN), Partial Least Squares Discriminant Analysis (PLSD), and SIMCA. The emphasis will be on using actual and synthetic data to gain a practical understanding of the different techniques and their proper application.
This short course is normally presented over 2 days. A 1-day version is also available, but we strongly recommend that you choose the 2-day version. The short course may be supplemented by additional workshop days during which the instructor works with your people on your actual applications and data.
This course will help the student understand:
-what the important discriminant techniques are and how they work;
-the steps necessary for the creation and successful deployment of calibrations;
-the similarity and differences among the techniques,
-the pitfalls and tradeoffs, and
-how to intelligently use them to improve calibrations and produce better analytical results;
-the strengths and weaknesses of each technique;
-how to select calibration standards, and
-how to assess the reliability of calibrations;
-which software capabilities and features are important for the student's applications.
This course will enable the student to:
-properly use the quantitative software provided with an instrument or commercial package;
-read critically and understand the current literature;
-further explore the topics using the course notes and comprehensive bibliography.
This course is intended for chemists, spectroscopists, chromatographers, biologists, programers, technicians, mathematicians, statisticians, managers, engineers, and anyone responsible for developing analytical calibrations using labortory or on-line instrumentation, managing the development or use of such calibrations and instrumentation, or designing or choosing software for the instrumentation. This introductory course requires no prior exposure to the material. Students who have explored the topics but are not yet comfortable using them will also find this course beneficial. The data-centric approach to the topics does not require any special mathematical background, but a familiarity with matrix multiplication would be helpful.
This course can be adjusted to meet your specific training requirements. The course can also be combined with a mentoring program and other consulting services to help your company quickly and productively achieve self-sufficiency.
Richard Kramer is President of Applied Chemometrics, Inc. and consults to instrumentation manufacturers and users. He is the author of the Chemometrics Toolbox software for use with MATLAB. His recent book, Chemometric Techniques for Quantitative Analysis has been listed as a bestseller in Chemistry. Since receiving his B.S. degree in chemistry from MIT, he has worked for over 15 years in analytical instrumentation and computer-based data analysis. Mr. Kramer has been instructing in the practical applications of chemometrics for over 10 years.