<b>Descriptive statistics and PLS calibration models for pol content of sugarcane juice samples extracted using the direct extraction method (DEM) and shredded extraction method (SEM)</b>

<p dir="ltr"><b>1. Descriptive statistics of the calibration and validation sets of sugarcane juice samples extracted using DEM and SEM</b></p><p dir="ltr">The calibration set and validation set of the individual extraction methods: DEM and SEM, as well as a combined calibration set (mixed), are presented below. The dataset of individual extraction method was ranked in descending order from the maximum value to the minimum value. Subsequently, the first three samples were assigned to the calibration set and the fourth sample was assigned to the validation set, resulting in a calibration-to-validation ratio of 75 % to 25 %, respectively. DEM calibration yielded an average pol content of 17.90 %, with the validation set yielding 17.88 %. Meanwhile, the SEM calibration set yielded an average pol content of 17.22 %, while the validation set yielded 17.20 %. The calibration sets of DEM and SEM were then combined to form the mixed calibration set, which yielded an average pol content of 17.56 %.</p><p><br></p><p dir="ltr">Table S1 Descriptive statistics of pol content in calibration and validation sets of sugarcane juice extracted using DEM and SEM</p><p dir="ltr">(EAEF180305_Table_S1.csv)</p><p dir="ltr"><br></p><p dir="ltr">DEM: direct extraction method, SEM: shredded extraction method, <i>n</i>: total number of samples, %: percentage of pol in sugarcane juice, <i>SD</i>: standard deviation, Min: minimum, Max: maximum.</p><p dir="ltr"><b>2. Performance of individual PLS calibration models for predicting pol content in sugarcane juice samples extracted using DEM and SEM, measured with a benchtop NIR spectrophotometer</b></p><p dir="ltr">Table S2 below presents unacceptable partial least squares (PLS) calibration models for pol content in sugarcane juice samples extracted using the direct extraction method (DEM) and the shredded extraction method (SEM) within selected wavelength ranges of 800–1100 nm and 1850–2500 nm. The DEM calibration model, when applied to predict DEM and SEM validation sets in these wavelength regions, produced higher root mean square error of predictions (<i>RMSEP</i>s) compared to the models developed over the wavelength range of 1100–1850 nm and 400–2500 nm. Similarly, the SEM calibration model yielded higher <i>RMSEP</i>s when used to predict pol content in both SEM and DEM validation sets. Furthermore, the second derivative (2D) preprocessed spectra for both the DEM and SEM resulted in higher <i>RMSEP</i>s than those obtained using raw spectra models. Therefore, these wavelength regions are considered unsuitable for predicting pol content in sugarcane juice samples.</p><p><br></p><p dir="ltr">Table S2 Performance of individual PLS calibration models for predicting pol content in sugarcane juice samples extracted using DEM and SEM, measured with a benchtop NIR spectrophotometer</p><p dir="ltr">(EAEF180305_Table_S2.csv)</p><p dir="ltr">DEM: direct extraction method, SEM: shredded extraction method, <i>n</i>: total number of samples, λ: wavelength, nm: nano meter, LV: latent variable, dash (-): the same calibration values of DEM and SEM are represented in SEM and DEM directly below the extraction treatment, 2D: second derivative preprocessed spectra, Cal: calibration, Val: validation, <i>R</i>_cv²: coefficient of determination of cross-validation, ^a<i>SEC</i>: root mean square error of calibration, <i>R</i>_p²: coefficient of determination of prediction, ^b<i>SEP</i>: root mean square error of prediction, %: percentage of pol content in sugarcane juice.</p>