Evaluation of Predictive Capability of Artificial Neural Network and Multiple Linear Regression: A Case Study of Lipid Extraction from Microalgae using Quaternary Solvents
No Thumbnail Available
Date
2019
Journal Title
Journal ISSN
Volume Title
Publisher
Published by Department of Civil Engineering University of Ilorin, Ilorin
Abstract
Abstract: Thiresent study was carried out to evaluate the performance of artificial neural network (ANN) and multiple linear regression (MLR) as modelling tools for predicting the lipid yield obtained from microalgae using quaternary solvent mixtures. The MLR models were developed using two different mathematical softwares viz. Microsoft excel (model 1) and Polymath (model 2). The augmented simplex lattice design under the mixture methodology of the Design Expert software was used to generate the design of experiments used in this study. A comparison of the models developed using ANN and MLR for the extraction process was carried out based on pertinent statistical parameters. Although the results from both MLR models were very close, the calculated values of coefficient of determination (R2) of 0.9934 and the average absolute deviation (AAD) of 2.4789 for the ANN model when compared with values obtained from the MLR (R2 = 0.9898 and AAD = 2.9821 for model 1) and (R2 = 0.9898 and AAD = 2.9825 for model 2) showed that the ANN model was more accurate and precise than the MLR models. The actual maximum lipid yield of 19.4 wt% lipid g-1 DCW was obtained at solvent mixture in the volume ratio of 1:5:1:1 for chloroform, methanol, ethanol, and dichloromethane respectively.
Description
Keywords
Artificial neural network, multiple linear regression, lipid extraction, modelling