Search or add a thesis

Advanced Search (Beta)
Home > Neural Networks Ensemble Evaluation of Aggregation Algorithms for Forecasting

Neural Networks Ensemble Evaluation of Aggregation Algorithms for Forecasting

Thesis Info

Access Option

External Link

Author

Saima Hassan

Supervisor

Jafreezal Jaafar

Department

Department of Computer and Information Science

Program

MS

Institute

Universiti Teknologi Petronas

Institute Type

Private

City

Seri Iskandar

Country

Malaysia

Degree Starting Year

2011

Degree End Year

2013

Viva Year

2013

Thesis Completing Year

2013

Thesis Completion Status

Completed

Page

108

Subject

Artificial Intelligence

Language

English

Link

https://www.researchgate.net/publication/313764198_Neural_Networks_Ensemble_Evaluation_of_Aggregation_Algorithms_for_Forecasting

Added

2021-02-17 19:49:13

Modified

2024-03-24 20:25:49

ARI ID

1676728085153

Similar


The aim of the thesis is to examine and analyze different aggregation algorithms to the forecasts obtained from individual neural network (NN) models in an ensemble. In this study an ensemble of 100 NN models are constructed with a heterogeneous architecture. The outputs from the individual NN models were combined by four different aggregation algorithms in NNs ensemble. These algorithms include equal weights combination of Best NN models, combination of trimmed forecasts, combination through Variance-Covariance method and Bayesian Model Averaging. The aggregation algorithms were employed on the forecasts obtained from all individual NN models as well as on a number of the best forecasts obtained from the best NN models. The output of the aggregation algorithms of NNs ensemble were analyzed and compared with each other and with the individual NN models used in NNs ensemble. The results of the aggregation algorithms of NNs ensemble are also compared with the Simple Averaging method. The performances of these aggregation algorithms of NNs ensemble were evaluated with the mean absolute percentage error and symmetric mean absolute percentage error. In the empirical analysis, the methodologies developed were tested on the Universiti Teknologi PETRONAS load data set of five years from 2006 to 2010 for forecasting. It can be concluded from the results that the aggregation algorithms of NNs ensemble can improve the accuracy of forecast than the individual NN models with a test data set. Furthermore, in the comparison with the Simple Averaging method, the aggregation algorithms of NNs ensemble demonstrate slightly better performance than the Simple Averaging. It has also been observed during the empirical analysis that; reducing the size of ensemble increases the diversity and, hence, accuracy. Moreover, it has been concluded that more benefits can be achieved by the utilization of an advanced method for forecast combinations.
Loading...

Similar Thesis

Showing 1 to 20 of 100 entries
TitleAuthorSupervisorDegreeInstitute
MS
Universiti Teknologi Petronas, Seri Iskandar, Malaysia
MS
Capital University of Science & Technology, Islamabad, Pakistan
REM
COMSATS University Islamabad, Islamabad, Pakistan
PhD
University of Engineering and Technology, Peshawar, Pakistan
University of Engineering and Technology, Rawalpindi, Pakistan
RCS
COMSATS University Islamabad, Islamabad, Pakistan
PhD
Ghulam Ishaq Khan Institute of Engineering Sciences and Technology, Swabi, Pakistan
PhD
Ghulam Ishaq Khan Institute of Engineering Sciences and Technology, Swabi, Pakistan
MSc
Quaid-i-Azam University, Islamabad, Pakistan
MSc
Quaid-i-Azam University, Islamabad, Pakistan
Mphil
Quaid-i-Azam University, Islamabad, Pakistan
RCS
COMSATS University Islamabad, Islamabad, Pakistan
RMS
COMSATS University Islamabad, Islamabad, Pakistan
MSc
Quaid-i-Azam University, Islamabad, Pakistan
MS
International Islamic University, Islamabad, Pakistan
MS
International Islamic University, Islamabad, Pakistan
MSc
Quaid-i-Azam University, Islamabad, Pakistan
Mphil
Quaid-i-Azam University, Islamabad, Pakistan
BS
COMSATS University Islamabad, Islamabad, Pakistan
MS
International Islamic University, Islamabad, Pakistan
TitleAuthorSupervisorDegreeInstitute
Showing 1 to 20 of 100 entries

Similar Books

Loading...

Similar Chapters

Loading...

Similar News

Loading...

Similar Articles

Loading...

Similar Article Headings

Loading...

بھٹو کا خطاب

بھٹو کا خطاب

چئیر مین بھٹو شہید شملہ معاہدے کے بعد عوام کے بہت بڑے اجتماع سے خطاب ۔نوے ہزار جنگی قیدیوں کی واپسی پانچ ہزار مربع کلو میٹر علاقہ واپس لیا جس پر رہنے والے دس لاکھ لوگ بے گھر ہو چکے تھے ان کی گھروں کو واپسی ہوئی لیکن جنہیں قید سے چھڑا کر لائے انہوں نے انعام میں تختہ دار کی زینت بنا دیا ۔

 

PENDIDIKAN ISLAM PADA MASA RASULLAH SAW. (PERIODE MEKAH DAN MADINAH)

Islamic education today cannot be separated from Islamic education in Islamic classical era. The Prophet Muhammad has served as a central figure of Islamic education from Islamic classical era to modern Era. The implementation of Islamic education in the time of the Prophet Muhammad can be categorized into Meccan period and Medina Period. In Meccan period, the prophet  put emphasis on tawhid, who used to adhare to politism, to adhare to monotism, that is to believe in Allah the only God. The strategy of education employed by the prophet was secret in nature. Initially, he conducated Islamic education amongst the members of his family and his companions then  to more extended cummunity. In Mecca, the Prophet made the house of al-Arqam ibn Abi Al-Arqam, as the centre of  Islamic education.  In Medinan period, the prophet conducted more complex  Islamic  education  than that  he did in Mecca. Islamic education conducted to covered  (a) Islamic brotherhood; (b) social walfare education;   and (c) nation defence education. In this period, it was mosque that served as the centre of Islamic education.

Utilizing Potential Oleaginous Microbes for Lipid Based Biofuel Production from Food Processing Waste

The adoption of alternative energy resources is essential in the current scenario of energy demand to reduce dependence on fossil fuels. The production of biofuels from oleaginous microorganisms is of increasing interest because of its potential benefits on plant-based biofuels. The sustainability of microbial biofuel and its complete independence from food crops make it an attractive option for biofuel production. However, one of the key barriers identified in the commercialization of biofuel from oleaginous microbes is to find a low-cost substrate to reduce the capital cost. In this study, the biofuel potential of bacterial lipids is successfully presented using food waste as a substrate. Oleaginous bacteria (n= 24) were isolated from crude oil and screened for their ability to store lipids using different types of food processing waste (apple, orange and mango waste) as a substrate. Five potent strains were used to test their tendency to reduce the pollution load in terms of removal of organic matter from different types of waste, in which strain KM9 and strain KM15 (Bacillus cereus) had the competent potential for degradation. The strain KM15 represented a greater potential for degradation of mango waste (volatile solids and COD removal up to 38.5% and 48.9%, respectively) while simultaneously exhibiting lipid accumulation up to 41.5% in 96 hours. Strain KM9 showed the highest potential for the treatment of orange waste with 47% removal of VS and 60% removal of COD. Further experiments confirmed the potential of KM9 to degrade citrus waste. The comparative study of strain KM9 with R. erythropolis (known oleaginous strain) showed that biomass production could be increased by supplementation of waste with nitrogen and carbon sources. xix These two strains also represented the potential for storing considerable lipids using waste as a substrate and modified limonene medium. GC-MS analysis revealed that both strains (KM9 and KM15) were able to store more saturated fatty acids (SFA) (38.4% and 39%, respectively) in their cells using a modified limonene medium compared to waste as substrate, however, the use of orange waste triggered the accumulation of monounsaturated fatty acids (MUFA). The high proportion of palmitic acid and stearic acid indicated its close resemblance to plant-based biofuels. The results of this study not only provide insights into the accumulation of bacterial lipids for sustainable biofuel production in order to meet the growing energy demand, but also offer an added benefit of minimizing waste.