PRS - back to the homepage   Pattern Recognition Systems - mail us {short description of image}
 
 
INFO
Software
| Courses
Consultancy
Order
News
Download
Support

10.04.00 mail uswebmaster

PRS as
MIX-Møhlenpris PB 24
5006 Bergen
NORWAY
Ph:
+47 5532 5221
Fax:
+47 5558 9496

Courses based on Sirius



PRS offer curses - both standard and special adapted

Regular Standard courses

PRS invite you to attend courses in multivariate analysis, experimental design and statistical process control. The courses are given by experts in the specific fields and you gain valuable insight at these courses. PRS offers both open courses and in-house courses. The majority of the courses are given in Norwegian or English. Courses can be provided on short notice throughout Europe and North America. Courses in the rest of the world might be possible. In addition we offer workshops and seminars

Each course is explained in more detail by following the links below.

Please observe that we have only listed the courses that we have already established as regularly events. We will offer several other courses, and we put them on our web pages a few months before they are arranged. Therefore we invite you to check our web-site regularly.

Courses based on use of Sirius

*Univariate and bivariate data analysis and regression
*Experimental design, optimisation and response modelling
*Multivariate problem solving
*Total Process Quality
----------------------------

Univariate and bivariate data analysis and regression


Course code: StatBasic

This short course aims at introducing new users to the basic terms and concepts in data analysis and statistics.

Background needed: None

The course starts with a lecture on the "problem chain", i.e. data, correlation, information and action. We then continue with a hands-on software tutorial on descriptive data analysis. The normal distribution and the use of normal probability plots is subsequently introduced as a mean to outlier detection. We proceed with the philosophy and principles of regression analysis.

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

Experimental design, optimisation and response modelling
Course code: ExpBasic

The course aims at learning you the basic principles of multivariate design and the main designs needed for practical experimentation.

Background needed: Elementary univariate statistics and regression corresponding to StatBasic.

The first day is devoted to basic experimental design. Factorial and fractional factorial designs represents the main theme. The use of normal probability plots for model selection and verification is highlighted with several practical examples. The day ends with a hands-on Plackett-Burnham screening design for selecting the important factors in a process step in the synthesis of a drug.

The second day starts with a hands-on software tutorial and continues with a lecture on the principles of optimisation and response modelling. Two hands-on software tutorials, one using central composite design (CCD) end the day. The use of contour plot for graphic presentation of a response surface is demonstrated.

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

Multivariate Problem Solving
Course code: MultBasic

This course aims at learning you practical problem solving by means of multivariate methods.

The course aims at an operational understanding of the concepts and methodologies collectively known as latent-variable analysis and regression.

Background needed: StatBasic or equivalent knowledge.

The first day starts with multivariate problem areas and shows how to adapt the strategy to the problem type. The philosophy and history of correlation and latent variable analysis follows. The connection between data, matrices and projections is revealed and principal component analysis (PCA) is introduced as the basic tool in exploratory multivariate analysis. Procedures for pre-treatment of data end the theoretical session, and the day is closed with a hands-on software tutorial illustrating different methods for exploratory data analysis.

On the second day we introduce the SIMCA method as a tool for classification of objects and revelation of discriminatory variables. Several examples and hands-on software tutorials represent the major part of the second day's content.

On the third day, the philosophy of principal components and partial-least-squares regression are introduced and a hands-on software tutorial combining mixture design and spectroscopic fingerprinting are used to demonstrate the use of PLS for multivariate calibration. We proceed with examples showing how to use mixture design and regression for optimising wine quality and the use of mixture design and PLS for on-line process control of feed. The course ends with a hands-on software tutorial on the use of PLS to predict octane number from infrared profiling.

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

Total Process Quality
Course code: TPQ

This course aims at learning you how to utilize multivariate methods in process exploration and optimisation and on-line process quality control.

Background needed: StatBasic or equivalent knowledge.

The first day starts with a review of the history and philosophy of univariate statistical process control. We continue with a lecture on Taguchi's quality philosophy and introduce the concept of total process quality. The usefulness of PCA and PLS for analysis of historic process data is demonstrated by means of a hands-on software tutorial on data from a factory. This example illustrates the potential of PCA and PLS for revealing critical process variables and lacking measurements.

The second day starts with an example of the use of PLS for building and validating a model for predicting the important quality characteristics octane number from infrared spectra. A lecture on the use of mixture design and PLS for rapid and cheap determination of feed composition in the manufacturing of polymers follows.

The day ends with a lecture on experimental design and a hands-on software tutorials screening for important factors.

----------------------------Top of Page
© 1997-04 PRS AS. All rights reserved.