By the analysis of the scores plot it was not possible to distinguish between banana and coconut samples based on their NIR spectra. In addition, the coffee samples were somewhat distant from the other groups in the scores plots. But, good calibration models have been already reported in the literature by combining coffee with banana biomasses [3]. Consequently, the above discussion justifies the combination of three different biomasses in a single calibration model, with the advantage of covering a wide range of variation and being as generic as possible.