| Chemometrics can be described simply as the application of statistical and mathematical techniques to the analysis of chemical data. For Near Infrared Spectroscopy (NIR), chemometrics extracts concentration information, product quality, identity or material grade from spectral data for samples of varying complexity. Chemometrics produces multivariate calibration models that can handle samples that have limited sample preparation, limited separation of components, multiple changing constituent levels and samples that do not exactly match those used to build the calibration. The processing power of modern computers along with user friendly software packages like TQ Analyst allow for quick and simple development of calibrations also called methods or models. Method development uses mathematical algorithms such as Partial Least Squares (PLS) to find patterns in spectra that explain the principal variation in the sample set. Principal Components are used to calculate scores which make a mathematical representation of spectra into a single point that can be plotted. These patterns can also be used to score samples spectrally based on how close they are to the average sample in the set. These plots or scores can be used to identify samples that are outliers, redundant and overall patterns in the samples. Multivariate techniques are typically used for NIR methods because of their advantages compared to univariate analysis such as traditional Beer’s Law analysis. Complete signal selectivity, finding the wavelengths that correspond to the analyte of interest, is not required in multivariate analysis. In addition multivariate analysis allows for calibration of multi-component samples in the presence of spectral interferents. Multivariate analysis also uses powerful diagnostic tools to aid in assessing the quality and reliability of the method on future spectral data. The application being tested often dictates which calibration modeling technique will work the best. |  | FAME calibration developed in
TQ Analyst for Biodiesel |
TQ Analyst (TQA) from Thermo Fisher Scientific makes chemometrics simple. TQ Analyst employs multiple modeling techniques for quantitative NIR analysis including Step-wise Multiple Linear Regression (SMLR), Partial Least Squares (PLS) and Principal Components Regression (PCR). Classification techniques in TQA include both spectral and principal component techniques like similarity match, distance match, discriminant analysis and soft independent modeling of class analogies (SIMCA). TQ Analyst uses a wizard which can guide you through all steps of method development. The wizard asks basic questions, analyzes spectra, determines method feasibility and suggests ways to improve the method. TQA menus and tabs are laid out to lead the user through a logical sequence of calibration development. If the user ever has a question about an option in TQA, they can click the explain button to pull up extensive help information on that topic. TQA is loaded with diagnostic tools, such as cross or independent validation, Principle component scores, predicted residual error sum of squares (PRESS), Residual plots and Loading spectra, that can help both novice and experienced method developers alike optimize their method. |
0 Items 
