Correlation Coefficient Matrix and Multiple Regression Analysis of Properties of Starches from Wild Plants in Japan

Toshihiko SUGANUMA, Shigeo FUJIMOTO, Kanefumi KITAHARA, Tomonori NAGAHAMA
1996 Journal of Applied Glycoscience  
Statistical analysis was done for the properties of 66 starches from wild plants in Japan. Correlation matrix of eight properties (numerical average size of granules, blue value, phosphorus content, enzyme digestibility, swelling power, solubility, gelatinization temperature, and maximum viscosity) showed a high positive correlation between the maximum viscosity and the swelling power (r =0.74), and high negative correlations between the size and the digestibility (r=-0.77) and between the
more » ... inization temperature and the swelling power (r= -0.73). Neither the phosphorus content nor the solubility has significant correlation coefficients for the, relationships with the others. Multiple regression analysis indicated that the maximum viscosity in amylogram is predictable to some extent from the variables of the granule size and the swelling power. Furthermore, with multiple regression equations obtained from the primary properties the enzymatic digestibility alone is predictable from the size and the blue value as major explanatory variables. Regression equations obtained by step-wise method suggest that the decrease (e.g. 0.1) in blue value by some genetic controls would bring about the improvement in digestibility (10% of the increase) and swelling power (two of the increase) of starches.
doi:10.11541/jag1994.43.355 fatcat:kfjehwgvyrbrnd3o2vhxjx2bcm