Abstract
The technique of near infrared reflectance spectroscopy was tested to predict the processing quality of potatoes by scanning (850–2500 nm) ground raw potatoes (n = 2517) from nine growing seasons and four to five locations. Calibration equations were developed for dry matter, starch, reducing sugars, sucrose and total sugar as well as lightness of dehydrated potatoes and potato crisps, and quality score of finished French fries. Two strategies in modelling were compared, an annually repeated (incremental) and a retrospective calculation at the end of the project with inclusion or removal of spectral outliers. The best coefficients of determination (r2) within independent validation sets were about 0.99 for dry matter, 0.96 for starch, 0.43 for reducing sugars, 0.71 for sucrose and 0.66 for total sugar. The RPD statistic (ratio of standard error of prediction to sample standard deviation) was up to 8.5 and 5.4 (dry matter and starch content, respectively), but was unacceptably low in the case of reducing sugars (1.7) and low for sucrose (2.4) and total sugar (2.3). Direct prediction of processing quality resulted in r2 values of 0.52 for dehydrated potatoes, 0.69 for potato crisps and 0.56 for French fries, but the RPD of these models were too low even for screening purposes (RPD < 2.0).
© 2011 IM Publications LLP
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription