Unified Discrete Wavelet Transform with Ridge Regression and Principal Component Regression to Predict Concentration of Gingerol Compound in Ginger Crop

Sony Sunaryo (Statistika, Institut Teknologi Sepuluh Nopember, Surabaya, INDONESIA)

IPTEK The Journal for Technology and Science
Vol 19, No 3 (2008)

Multivariate calibration model can be used as an alternative method to predict the concentration of a gingerol compound. The prediction usually are carried out chemically through a long and expensive process using High Performance Liquid Chromatography (HPLC) measurement method. Since the number of samples (n) are less than of the number of independent variables (p), and between the independent variables are correlated, so the development of model using conventional regression is not valid. The combination of Discrete Wavelete Transform (DWT) with Ridge Regression and Principal Component Regression have been adopted in this paper to predict concentration of gingerol, and it showed a promising result.