The GRAMS IQ application records the training sample information and lets users enter custom information as needed. Easy-to-use tools allow the creation of multiple combinations of modeling conditions—a series of separate “experiments” that completely define the model to be calculated.

With calculations performed, users view and plot all calculated data as well as a variety of diagnostic indicators stored in the experiments. Results of the various calibration approaches can be compared using built-in or user-defined data analysis plots. With the optimum conditions determined, users can save one or more calibrations to a file for use in prediction.

All calculated data and graphs export to a rich text document for quick and easy reports.

• Organizes calculated data with a copy of original raw calibration data into one file
• Keeps a history of all attempted calibration models and easily clones experiments to try multiple variations of similar conditions
• Includes multivariate algorithms as well as qualitative methods for simple data exploration and visualization
• Lets users construct complete models of spectra using a PCA/MDR discriminant analysis algorithm without any constituent information

• Can remove undesirable variations from data before it is used in any calculations
• Offers click-and-drag region selection; the visual spectral region editor can superimpose data set variance or constituent correlation spectra over raw or processed spectra for optimal region selection
• Lets users simultaneously view and overlay multiple plots and diagnostic indicators from one or more experiments, simplifying the process of model selection/optimization

• Familiar, easy-to-use editor accepts data from typed input or Windows™ Clipboard
• Software imports files from previous versions of PLSplus and Discriminate as well as data sets with constituent information from other vendor formats
• Multiple calibrations from different training sets can be saved in the same file to create a complete collection of all analyses to be performed on each sample spectrum and linked indelibly to the training data file used to create them for audit purposes

• Lets users specify how spectra are to be analyzed and how to output results
• Allows users to load one or more spectra and predict them against any GRAMS IQ calibration model, helping to determine if the spectrum matches the calibration well enough for an accurate prediction

• Can customize prediction report to include constituent quantities, Mahalanobis distance, model limits tests and spectral residual
• Can be configured to operate in “auto-predict” mode, which provides a simplified interface ideal for routine analyses, repetitive measurements, quality control and fieldwork