Development of a Low-Cost System for Liquid Clustering Using a Spectrophotometry Technique

Abstract

The necessity for point-of-care, low-cost devices for early screening are an important issue at hand. Several methods to identify liquids using their spectrum have been analyzed and liquids with small differences in their molecules have been identified. Methods of dimensionality reduction, such as principal component analysis, are used to check the clustering of different liquids. In this paper, an optical instrumentation development is approached, using six wavelength values of the visible light spectrum, to identify six different liquid samples, urine, drinking water, vanilla flavoring liquid, Surabaya’s tap water, and yellow food color. After a normalization process and by using a principal component analysis dimensionality reduction from six to two dimensions, 97.61 percent of the information was captured, and the system was able to differentiate all five samples into different clusters.