scholarly journals Prediction of Daily Mean PM10 Concentrations Using Random Forest, CART Ensemble and Bagging Stacked by MARS

2022 ◽  
Vol 14 (2) ◽  
pp. 798
Author(s):  
Snezhana Gocheva-Ilieva ◽  
Atanas Ivanov ◽  
Maya Stoimenova-Minova

A novel framework for stacked regression based on machine learning was developed to predict the daily average concentrations of particulate matter (PM10), one of Bulgaria’s primary health concerns. The measurements of nine meteorological parameters were introduced as independent variables. The goal was to carefully study a limited number of initial predictors and extract stochastic information from them to build an extended set of data that allowed the creation of highly efficient predictive models. Four base models using random forest, CART ensemble and bagging, and their rotation variants, were built and evaluated. The heterogeneity of these base models was achieved by introducing five types of diversities, including a new simplified selective ensemble algorithm. The predictions from the four base models were then used as predictors in multivariate adaptive regression splines (MARS) models. All models were statistically tested using out-of-bag or with 5-fold and 10-fold cross-validation. In addition, a variable importance analysis was conducted. The proposed framework was used for short-term forecasting of out-of-sample data for seven days. It was shown that the stacked models outperformed all single base models. An index of agreement IA = 0.986 and a coefficient of determination of about 95% were achieved.

Author(s):  
Mohammed Okoe Alhassan ◽  
Michael Boakye Osei

Soft-computing techniques for fire safety parameter predictions in flammability studies are essential for describing a material fire behaviour. This study proposed, two novel Artificial Intelligence developed models, Multivariate Adaptive Regression Splines (MARS) and Random Forest (RF) methods, to model and predict peak heat release rate (pHRR) of Polymethyl methacrylate (PMMA) from Microscale Combustion Calorimetry (MCC) experiment. From the statistical analysis, MARS presented the highest coefficient of determination (R2) values of (0.9998) and (0.9996) for training and testing respectively, with low MAD, MAPE and RMSE values. Comparatively, MARS outperformed RF in the predictions of pHRR, through its model algorithms that generated optimized equations for pHRR predictions, covering all non-linearity points of the experimental data. Amongst the input variables (sample mass, THR, HRC, pTemp and pTime), heating rate (β), highly influenced pHRR outcome predictions from MARS and RF models. However, to validate the performance and applicability of the proposed models. Results of MARS and RF were benchmarked with that from Artificial Neural Network (ANN) methods. The MARS and RF models observed the least error deviation when compared with pHRR results for PMMA from the ANN models. This study therefore, recommends the adoption of MARS and RF in the predictions of flammability characteristics of polymeric materials.


Sign in / Sign up

Export Citation Format

Share Document