Chemometrics : Quality Assurance and Control Process
| Unscrambler Online | ||||||||
 Useful Links
|
It is of great advantage for scientists to use spectroscopic and chromatography instruments to measure product quality. However, quality is difficult to quantify. It takes years of experience to acquire the skills to ensure that a product’s flavors meet standards, identify process changes, and detect ingredient changes by observing color, texture, taste, aroma, etc. Using CAMO’s software products for chemometrics, offers a faster and more precise assessment of composition, physical, and sensory properties of a product. In analyzing multivariate data with our software, patterns emerge that are related to product quality. The data patterns are correlated and used to develop models that predict quality parameters for future data. These models are repeatable spectral profiles that can be applied to every batch sent to market. With CAMO statistical software you can automate quality control to find new patterns in the data.
Sensory and Research Scientists need quick results that pay dividends. They need new and better ways to use data in more efficient ways, and multiple options to analyze and alternate between numbers of compounds simultaneously. CAMO’s statistical software offers superb graphics for plotting, time-series analysis and multilevel designs.
CAMO multivariate statistical software products and solutions functionalities are excellent in the areas of Chemometrics / Chemistry, Spectroscopy, Screening, Quality Control, Design of Experiments and Product Development.
Steps in the Quality Assurance and Control Process:
Explorative analysis:
Discover sample patterns, hidden phenomena, and variable relationships in your data that visual inspection or traditional statistical methods cannot reveal. The method used for this is Principal Components Analysis (PCA), but you also have traditional statistical methods available.
Regression analysis:
In many cases you want to relate the hidden phenomena discovered in your data to specific responses, properties, or maybe quality parameters of your "system". This leads to a so-called calibration or regression model. The available regression methods are Multiple Linear Regression (MLR), Principal Components Regression (PCR), Partial Least Squares 1 (PLS1), and Partial Least Squares 2 (PLS2).
Prediction and validation:
Once you have related the hidden phenomena to your chosen responses, you can actually use the calibration model to predict (estimate) the future responses of new samples.
Classification (SIMCA):
Instead of relating hidden phenomena to responses or properties, you can also use them to identify groups or classes of samples with similar hidden phenomena.
Design of Experiments:
It is important to get representative data for the analysis. Experimental design makes sure the samples contain information. There are a number of screening and optimization designs available. Analysis of designs is done using ANOVA and Response Surface Analysis. It is also possible to analyze experimental designs using the methods above. Designs include Fractional and Factorial, Placket-Burmann, Box Behnken, Central Composite, Classical mixture designs, and D-optimal designs for mixtures and non-mixtures.
CAMO multivariate products and solutions have been used for Quality Assurance and Quality Control, in the following verticals:
- Food & Beverage
- Agriculture, Soil and Environment
- Oil and Gas
- Chemicals, Polymers and Manufacturing
- Paper and Wood
- Pharmaceutical and Biotech
- Healthcare services
- Personal Household Products
Example
The Unscrambler® used in Quality Assurance and Control Process for Petroleum and Chemical Research
Industry uses for Petroleum and Chemical Research
The Unscrambler has provided the ability for our laboratory to facilitate the construction and deployment of discriminant and quantitative prediction models in record time with significantly shorter learning curves relative to spectroscopy-centric third-party applications… ”
Industry Specific uses
Research Scientists need quick results that pay dividends, new and better ways to use data in more efficient ways, and multiple options to analyze and alternate between numbers of mixtures simultaneously. Furthermore, compatible data formats and exports for use with instruments allow for better and easier data interfacing. The Unscrambler’s statistical functionalities are excellent in the areas of Chemometrics/ Chemistry, Spectroscopy, Screening, Quality Control, Design of Experiments, Product Development, and Instruwwments Interfacing.
Chemometric and Spectroscopy Uses:
NIR, MIR, FTIR, Raman Techniques, SPE, Sample Preparation, Chromatography, Octane Rating of Gasoline, Analysis of Petroleum Mixtures, Experimental Sample Analysis
