New structure-based models for the prediction of flash point and autoignition temperatures of alkyl esters
Created by W.Langdon from
gp-bibliography.bib Revision:1.8010
- @Article{HEIDARI:2023:chemolab,
-
author = "Zeinab Heidari and Mohammad Amin Sobati",
-
title = "New structure-based models for the prediction of flash
point and autoignition temperatures of alkyl esters",
-
journal = "Chemometrics and Intelligent Laboratory Systems",
-
volume = "240",
-
pages = "104877",
-
year = "2023",
-
ISSN = "0169-7439",
-
DOI = "doi:10.1016/j.chemolab.2023.104877",
-
URL = "https://www.sciencedirect.com/science/article/pii/S0169743923001272",
-
keywords = "genetic algorithms, genetic programming, Alkyl esters,
Enhanced replacement method (ERM), QSPR, Genetic
programming (GP), Support vector regression (SVR),
Random forest regression (RFR)",
-
abstract = "In this study, new models based on quantitative
structure-property relationship (QSPR) have been
proposed for the prediction of autoignition temperature
(AIT) and flash point (FP) of different alkyl esters.
In this regard, data sets containing 126 and 179 alkyl
esters from 10 categories were used for AIT and FP,
respectively. The enhanced replacement method (ERM) was
applied for choosing the appropriate molecular
descriptors, and linear models were developed based on
the selected descriptors. Nonlinear models were also
developed for AIT and FP using genetic programming
(GP), support vector regression (SVR), and random
forest regression (RFR). Then, the predictive
performance of each model was evaluated through
internal and external validation techniques considering
several statistical parameters such as coefficient of
determination (R2), root mean square error (RMSE), and
percent of average absolute relative deviation (AARD).
The outcome of the validation techniques confirmed the
satisfactory agreement between the predicted and
experimental data. The AARDpercent, R2, and RMSE for
the best corresponding models (GP-based nonlinear
models) was reported as 4.54percent, 0.74, and 39.01
for AIT and 2.51percent, 0.98, and 12.32 for FP,
respectively. The superiority of the new models over
the previous models was also proved by comparing the
prediction capability of different models",
- }
Genetic Programming entries for
Zeinab Heidari
Mohammad Amin Sobati
Citations