Classification and prediction of maize hardness-associated properties using multivariate statistical analyses Academic Article uri icon

abstract

  • Maize kernel hardness and proximate constituents were evaluated by several methods. Proximate constituents and hardness-associated properties were significantly correlated, and their correlation coefficients were improved when moisture, protein, and oil contents were kept constant. Multivariate techniques were applied to create new sets of variables to characterize maize hardness. The principal component scores created by principal component analysis were subjected to cluster analysis and discriminant analysis. Three factors interpreted as representing physical and chemical properties were identified using factor analysis. A total of 248 maize samples were grouped into 7 and 10 subgroups by cluster analysis. Beale's pseudo F statistic and the plot of principal component scores showed that 10 clusters would be a better solution than 7 clusters. The groups resulting from cluster analysis seem to have unique physical and chemical properties showing the different order in values of hardness measurements. Thus, the grouping by cluster analysis according to hardness-associated physical and chemical properties improved the explanation of maize kernel hardness. The application of discriminant analysis to create a classification rule for hardness clusters revealed that new observations had an 87% correct classification into hardness clusters. 2004 Elsevier Ltd. All rights reserved.

published proceedings

  • Journal of Cereal Science

author list (cited authors)

  • Lee, K., Herrman, T. J., Lingenfelser, J., & Jackson, D. S.

citation count

  • 35

complete list of authors

  • Lee, Kyung-Min||Herrman, Timothy J||Lingenfelser, Jane||Jackson, David S

publication date

  • January 2005