| Summary: | In this study, Multiple Linear Regression (MLR) was used to predict Brix
and pH on climacteric fruit i.e. bananas and tomatoes as well as non-climacteric
fruit i.e. strawberry and lime based on RGB and the La*b* color value. Fruits
used in this study were in various maturity stage from unripe to ripen. RGB and
La*b* color parameters were measured non-destructively using colormeter,
meanwhile, the internal quality measurements such as Brix and pH were
determined destructively or by conventional procedures in the laboratory. The
Unscrambler ® X 10.3 (CAMO, U.S., OLSO, Norway, the trial version) was used
for multivariate analysis. The accuracy of the statistics used in selecting the model
MLR was the correlation coefficient (r), Standard Error of Prediction (SEP), and
Bias. The minimum limit of the correlation coefficient should be greater than 0.5
(r> 0.5), followed by SEP and bias were small. Result showed that MLR
calibration model resulted in good calibration model based on RGB and La*b* for
banana, but less satisfactory calibration model for tomato, strawberry, and lime.
Therefore, validation model was only used for banana using different samples.
The relationship between the actual and predicted values of the MLR models was
determined from R2 (coefficient of determination). The best model for predicting
Brix and pH of banana based on La*b* color values resulted in the coefficient
determination (R2) between actual and prediction values of 0.78 and 0.61 for
validation model.
Keywords: RGB, La*b*, Brix, pH, multivariate analysis, Multiple Linear
Regression (MLR)
|
|---|