297 Performance expectations for near infrared analyzers.

C. R. HURBURGH, Jr. (1), G. R. Rippke (1), and T. J. Brumm (2). (1) Agricultural and Biosystems Engineering, Iowa State University, Ames, IA 50011; (2) MBS, Inc., Story City, IA 50248.

As near-infrared analysis becomes more widespread in trading and in other general applications, there is a need for criteria on which to evaluate performance. There are four areas of interest in evaluating near- infrared results: 1) accuracy relative to reference data; 2) precision (repeatability) among repeat analyses; 3) standardization across like units; and 4) sensitivity to temperature variations and other non-standard conditions. Data from four years’ experience with corn (moisture, protein, oil, starch, density) and soybeans (moisture, protein, oil) in 150 near-infrared transmission analyzers of two brands is used to support achievable performance targets for these areas. Targets are expressed in percentage of original validation standard error of prediction (SEP) relative to references. Two of the more important targets are for transferability to be 50% or less of SEP and for precision to be 33% or less of SEP. Approximately 10,000 data points, collected during both initial standardization and subsequent on-site quality control, were used to develop the criteria.