KNN algorithm and multivariate analysis to select and classify starch films

Abstract

Biodegradable starch films are promising as primary food packaging, and the k-Nearest Neighbor (KNN) algorithm enables selection and classification according to pre-established parameters. Here, the KNN algorithm and principal component analysis prove to be useful tools for sorting and selecting biodegradable starch packaging. Twelve biodegradable films were produced using starch from different botanical sources by the casting method. The KNN analysis evaluated data on thickness, water vapor permeability, tensile strength, elongation, water activity, transparency, and opacity, to obtain an information bank with 36 samples. Biodegradable films are visually homogeneous, transparent, without deformation, and easy to handle. The formulation (Cassava 5%) was classified as the best film, with WVP 1.21 × 10−10 (g. m−1. s−1. Pa−1), TS 2.34 (MPa), thickness 0.193 (mm), Aw 0.408, transparency 0.55 and opacity 0.63. The KNN algorithm and principal component analysis are advanced tools for classifying and selecting biodegradable starch films.