Goals for General Training Course
- Introduction to basic ideas in multivariate statistics as applied to chemical data
- Description of computations underlying several popular algorithms
- Hands on training with Pirouette
- Learn the important steps in data analysis: problem definition, data collection, data quality verification, and model creation and deployment
Multivariate analysis is a vast and complicated topic that cannot be explained in three days. To help attendees establish a foothold and feel comfortable asking the many questions required to eventually gain mastery in the area, we employ several approaches: lecture, demonstration and lab. Lectures, mostly focused PowerPoint presentations are followed by demonstrations inside Pirouette and then with supplied written exercises, which emphasize core concepts and provide attendees with a chance to use Pirouette in a realistic way. Attendees are also encouraged to send in their own data for class display and discussion and for overall enhancement to the course. Attendees are asked to send the data prior to the course so that data can be analyzed and prepped for use in class.
At the end of the course, attendees will be able to:
- Understand the need for exploratory data analysis
- Organize data for pre-processing, visualization and feasibility for various modeling algorithms
- Discover similarities or groupings
- Construct predictive models
- Deconvolve complex mixtures
- Categorize unknown samples
- Predict physical properties
- Grasp importance of issues in multivariate analysis: data scaling, overfitting, outlier diagnostics, transfer of calibration, etc.
- Achieve effective multivariate calibration
Below is an outline and sample schedule of what attendees can expect for the three day course.
- Overview of course topics and schedule
- Visual data analysis. Mixture analysis
- Introduction to Pirouette: loading files, interactions with graphics, and running algorithms.
- Discussion of principal component analysis (PCA), a powerful technique integral to any multivariate investigation
- Live demonstration of PCA focusing on meaning of the many results produced by the algorithm
- Hands-on exercises for attendees
- Discussion of pattern recognition concepts and methods
- Live demonstrations of KNN and SIMCA
- Hands-on exercises: focus on pattern recognition results and interpretation
- Discussion of multivariate regression, including background of PLS and PCR
- Live demonstrations of regression algorithms
- Final hands-on exercises: focus on regression methods and interpretation
- Advanced topics, depending on the specific interests of the attendees. Topics may include: chromatographic alignment, robust outlier detection, automating predictions, and interaction of Pirouette components with databases (Bio-Rad’s KnowItAll, Northwest Analytics, Quality Analyst, Access and SQL).
Goals for Advanced Training Course
Infometrix has been developing approaches devoted to optimizing and maintaining spectroscopic model performance in support of refinery and chemical plant process monitoring. We have examined a series of algorithmic approaches with the goal of streamlining the process of chemometric model construction to make the models significantly more robust when put into routine practice. This effort has generated the following observations:
- Even though there are published “Best Practices” for generating chemometric models, these practices are infrequently followed;
- Recalibration of an optical spectrometer may be perceived as warranted due to changes in crude slates and blending component composition, but may not be required;
- Even if a calibration was performed properly during initial installation, staffing changes and lack of training undermine subsequent recalibrations; and
- It is of benefit to minimize software maintenance frequency to control product giveaway.
In order to use optical analyzers effectively, the analyst needs to consider limitations that ultimately determine the eventual quality and life-expectancy of any calibration. There are many areas that have an impact on a chemometrics calibration: outlier detection, data treatments, selection of the number of factors, and even choice of algorithm.
Our new Advanced Spectroscopic Techniques course will cover strategies that lead to an improvement in calibration procedures for on-line and laboratory multi-wavelength spectrometers. In the course, attendees will be offered a thorough examination of important principles and will be guided through calibration strategies with a hands-on approach, resulting, hopefully, in a good understanding of potential improvements to spectroscopic classification, prediction, and long-term model reliability.
Goals for Custom Training Course
Contact us if you have questions or requirements that are not outlined here and we can organize courses specific to the needs of your organization.
|3-day general training course||Generally held at the customer’s location, this course constitutes a three-day, hands-on, intensive introduction to chemometrics.|
|Custom training course||One to five day courses can be offered according to the requirements of the customer. This option grades into one-on-one consulting or custom software development.|
|2-day advanced training course||Generally held at the customer’s location, this course constitutes a two-day, hands-on, intensive advance course focused only on calibration.|
In addition to our chemometric classes, we also offer custom software development and other special programs. Contact us for questions, quote or additional information.