For many years now, Multivariate Analysis (MVA) has been used by spectroscopists, analytical chemists, process engineers and sensory scientists to find important relationships in complex data tables. Methods such as Principal Component Analysis (PCA) have been used for data mining and exploratory data analysis purposes, while methods such as Partial Least Squares (PLS) have been used for predicting difficult to measure properties of products ranging from pharmaceuticals, petrochemical and agricultural comodities.
CAMO Software has been providing world recognized training courses in MVA for many years and this course on Advanced Multivariate Analysis builds on concepts introduced in other courses currently offered.
The course is a balance of spectroscopic and non-spectroscopic data and also touches on the concept of Data Fusion, an important area of MVA where multiple data sources can be aligned and advanced models developed.
All course material will be supplemented with hands on practical exercises that highlight the use of each method discussed in a practical manner.
Who should participate in this program?
It is targeted towards those practitioners who have a fundamental understanding of PCA and PLS and a basic understanding of preprocessing, who need a more in depth knowledge of the models they develop on a regular basis.
- A short overview of PCA and PLS: Setting the scene
- Some advanced data preprocessing options for spectroscopic and non-spectroscopic data
- PCA in depth,
- Utilizing PCA Diagnostics tools
- Model Stability and the use of objective cross validation
- PCA Projection and its importance in screening data before multivariate prediction
- PCA for Classification
- An introduction to Principal Component Regression (PCR)
- PLS in depth,
- Interpreting Loadings and Loading Weights in a joint manner
- Outlier tools and statistics for improving PLS models
- Inlier Statistics for detecting new samples for model updating during prediction
- Multivariate Model Maintenance Roadmap
- An Introduction to Support Vector Machines for Classification and Regression
- Hierarchical Modelling for developing complete decision making systems
- Conclusion and Summary
Dr. Heather Brooke Chief Chemometrician
Heather Brooke began her interest in chemistry at a young age when she took a chemistry mini-course in 3rd grade. She later earned a BS in Chemistry from Lander University in 2002. Shortly after graduating, she obtained a position as a lab technician at Fujifilm in Greenwood, SC. After working there for 3 years, with both chemists and engineers, she decided to apply herself to a new challenge: graduate school.
Heather worked under Dr. Michael Myrick as a Copenhaver Fellow at the University of South Carolina, earning a PhD in Analytical/Physical Chemistry in 2010. She gained teaching experience by working as a lab instructor at Lander University for Consumer Chemistry and at USC for both General and Physical Chemistry labs.
Heather´s work in Infrared Imaging has been recognized in the spectroscopy community since 2009, and was featured on CNN´s Big I and NPR´s Science Friday. She continued her research in analytical chemistry as a postdoctoral fellow at the US Naval Research Laboratory through the NRC/RAP postdoctoral fellowship program with a focus on data fusion. She then spent 2 years working in the PAT group at Merck. She is happy to have joined the team at CAMO software, and is looking forward to getting back into the classroom.
The course fees includes lunch, tea / coffee, course material and 30-days trial installation of The Unscrambler. Participants are required to bring their own Laptop, organize transport to / from training venue and accommodation at their own cost. Deadline for registrations: 2 weeks before the course start.
30 days net from date of invoice.
Cancellations up to 1 week prior the course start date, will be refunded with 50% of the registration fee, after this limit will not be refunded.
The course participants may be substituted or join a later training provided that CAMO is notified.