Application of linear regression algorithm and stochastic gradient descent in a machine‐learning environment for predicting biomass higher heating value

• Published in: Biofuels, bioproducts and biorefining Vol. 14; no. 6; pp. 1286 - 1295
• Main Authors: Ighalo, Joshua O., Adeniyi, Adewale George, Marques, Gonçalo
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.11.2020; Wiley Subscription Services, Inc
• Subjects: Algorithms; Artificial intelligence; Biomass; Calorific value; Data analysis; Heating; HHV; Learning algorithms; linear regression; Machine learning; Regression analysis; Root-mean-square errors; School environment; stochastic gradient descent; Stochasticity
• ISSN: 1932-104X; 1932-1031
• DOI: 10.1002/bbb.2140

The higher heating value (HHV) provides information about the quantity of energy contained in a fuel such as biomass. Correlations and models can be developed to predict biomass HHV quickly from other analysis data. In this study, a linear regression algorithm (LRA) and stochastic gradient descent (SGD) in a machine‐learning environment were used as novel methods to predict the HHV of biomass. The basis of the model was 78 lines of combined proximate and ultimate analysis data. The LRA model was observed to be more accurate. The testing for both models was done by stratified cross‐validation, stratified shuffle splits, and no sampling tests. The root mean square error (RMSE) of the LRA and SGD models was 8.151 and 21.65 kJ kg−1 and the mean absolute error (MAE) was 6.823 and 13.87 for the stratified shuffle split (ten random samples with 75% data). The coefficient of determination for both models was >0.999 in all cases. The study observed that LRA and SGD are among the most accurate artificial intelligence models for the prediction of biomass HHV. © 2020 Society of Chemical Industry and John Wiley & Sons, Ltd