On-site analysis system with central processor and method of analyzing

A method for developing a calibration model for use in a central processor configured to generate a predicted value of a property of interest based on an input from one or more data acquisition instruments, comprising:

• obtaining a set of samples having the property of interest;
• generating a first set of data by measuring the property of interest in the samples;
• generating a mathematical model of the property of interest using the first set of data;
• generating predicted values for the property of interest using the mathematical model;
• determining whether the predicted values for the property of interest correlate with the measured values of the property of interest to within a predetermined criterion;
• identifying the type and range of a secondary variable that is potentially influential of the measurement of the property of interest;
• determining whether the secondary variable is actually influential of the measurement of the property of interest;
• adjusting the preliminary model to compensate for the secondary variable and/or identifying an appropriate method of treating responses of the one or more data acquisition instruments to compensate for the secondary variable;
• determining whether predicted values of the property of interest generated by the adjusted mathematical model and/or using the treatment method correlate with the measured values within another predetermined criterion when the secondary variable is varied;
• generating a second set of data by measuring the property of interest in the samples using the treatment method;
• generating a prediction of the property of interest using the second set of data and the revised property model;
• determining whether a probable outlier is present in the prediction of the property of interest;
• generating a third set of data using the pretreatment method while varying the influential factor;
• generating a prediction of the property of interest using the revised property model and the third set of data;
• and
• determining whether a probable outlier is present in the prediction of the property of interest generated using the revised property model.