Optimized Extreme Learning Machine (ELM) Based on Genetic Algorithm (GA) To Predict Carbon Price Under The Influence of Multiple Factors

doi:10.21203/rs.3.rs-702953/v1

Download PDF

Research Article

Optimized Extreme Learning Machine (ELM) Based on Genetic Algorithm (GA) To Predict Carbon Price Under The Influence of Multiple Factors

https://doi.org/10.21203/rs.3.rs-702953/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

The promotion of carbon market can accelerate the pace of low-carbon transformation of China's economic structure and achieve more efficient carbon emission reduction. Accurate carbon price prediction is conducive to improving the risk management of carbon market and the decision-making of investors, but it also brings great challenges to relevant industry practitioners and the government. In this paper, a new hybrid model is proposed, which combines complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and genetic algorithm (GA) optimized extreme learning machine (ELM). The application of GA-ELM in carbon price prediction is firstly studied in this paper. Eight intrinsic mode functions and one residual can be obtained by CEEMDAN, and then partial autocorrelation (PACF) is used to determine the partial correlation between each sequence and its lag data, and they were taken as internal factors affecting the prediction. At the same time, energy, economic and social factors are selected as the external factors affecting the prediction, and the carbon price prediction is realized through internal and external factors. It has been proved that the model successfully overcomes the challenge of carbon price prediction based on multiple influencing factors. The hybrid model shows superiority in Beijing, Shanghai and Guangdong. The results show that the prediction performance of the proposed model is the best among the 15 models, and the prediction accuracy will be improved due to the decomposition of the carbon price. Besides, the CEEMDAN-GA-ELM model better overcomes the challenge of carbon price prediction with multiple influencing factors. This model provides a novel and effective tool for the government and enterprises to predict the carbon price.

Environmental Engineering

Environmental Policy

Carbon price forecasting

Complete Ensemble Empirical Mode Decomposition with Adaptive Noise

Extreme Learning Machine Optimized by Genetic Algorithm

Multiple factors affect prediction

Table 1 Abbreviations in this paper

Abbreviations	Definition
CEEMDAN	Complete ensemble empirical mode decomposition with adaptive noise
GA	Genetic algorithm
ELM	Extreme learning machine
GA-ELM	Extreme learning machine optimized by genetic algorithm
GRA	Grey correlation analysis
PACF	Partial autocorrelation function
EMD	Empirical mode decomposition
EEMD	Ensemble empirical mode decomposition
IMF	Intrinsic mode function
BP	BP neural network optimized by particle swarm optimization algorithm
PSOBP	BP neural network optimized by particle swarm optimization algorithm optimized by particle swarm optimization algorithm
LSSVM	Least squares support vector machine
	Goodness of fit
RMSE	Root mean square error
MAPE	Mean absolute percentage error
	Common factor 1 extracted by Factor analysis
	Common factor 1 extracted by Factor analysis

Table 2 Data sources

	source
The carbon price	China Emissions Trading Website（http://www.tanpaifang.com/）
The price of coal	China zhengzhou Commodity Exchange（http://www.czce.com.cn/）
WTI crude oil prices	Yingwei Caiqing Financial Information Portal（https://cn.investing.com/）
Futures price of natural gas	Yingwei Caiqing Financial Information Portal（https://cn.investing.com/）
The exchange rate	Fred Economic Data（https://fred.stlouisfed.org/）
Carbon neutral	Baidu index（http://index.baidu.com/v2/index.html#/）
Carbon trading	Baidu index（http://index.baidu.com/v2/index.html#/）
Climate change,	Baidu index（http://index.baidu.com/v2/index.html#/）
Carbon footprint	Baidu index（http://index.baidu.com/v2/index.html#/）
Carbon sequestration	Baidu index（http://index.baidu.com/v2/index.html#/）

Table 3 Details of data from three carbon markets

Market	Beijing	Guangdong	Shanghai
Tape	BEA	GDEA	SHEA
Data	2019/1/9~2019/12/31
The case number	239	239	239
Mean standard error	0.63966	0.16971	0.26453
The standard deviation	9.88897	2.62364	4.08957
The variance	97.792	6.883	16.725
Range	39.10	10.11	20.68
The minimum value	48.40	17.90	27.32
The maximum	87.50	28.01	48.00

Table 4 Correlation analysis of 9 influencing factors of carbon price.

Influencing factors	The correlation	Influencing factors	The correlation
The exchange rate	0.9451	Climate change	0.8864
The price of coal	0.9407	Futures price of natural gas	0.8786
WTI crude oil prices	0.9364	Carbon trading	0.8618
Carbon sequestration	0.9065	Carbon neutral	0.7938
Carbon footprint	0.9038

Table 5 KMO and Bartlett test.

KMO	0.774
Bartlett sphericity test	Approximate chi-square	2460.058
	Degree of freedom	10
	Significance	0.000

Table 6 The results of factor analysis

Influencing factors	Variance of common factor	Component score coefficient		Cumulative contribution(%)
Influencing factors	Variance of common factor	Factor 1	Factor 2	85.804
Climate change	0.864	-0.266	0.642
Carbon sequestration	0.861	0.230	0.166
Carbon trading	0.886	0.475	-0.189
Carbon neutral	0.875	0.469	-0.183
Carbon footprint	0.805	-0.053	0.459

Due to technical limitations, table 7 is only available as a download in the Supplemental Files section.

Table 8 Predictive effects of all models(Beijing)

The model number
1	BP	0.6883	5.9143	0.0687
2	LSSVM	0.7580	5.0571	0.0518
3	PSO-BP	0.7721	4.5694	0.0490
4	ELM	0.8139	4.3848	0.0442
5	GA-ELM	0.8428	4.1994	0.0402
6	EEMD-ELM	0.8543	4.1761	0.0486
7	EEMD-BP	0.9438	2.5927	0.0295
8	EEMD-PSO-BP	0.9636	2.1347	0.0248
9	EEMD-LSSVM	0.9796	1.5619	0.0177
10	EEMD-GA-ELM	0.9890	1.1473	0.0127
11	CEEMDAN-ELM	0.9074	3.4077	0.0347
12	CEEMDAN-BP	0.9503	2.4964	0.0274
13	CEEMDAN-PSO-BP	0.9759	1.6993	0.0179
14	CEEMDAN-LSSVM	0.9831	1.4568	0.0169
15	CEEMDAN-GA-ELM	0.9898	1.1294	0.0120

Table 9 Predictive effects of all models(Shanghai and Guangdong)

Model	Shanghai			Guangdong
Model
EEMD-ELM	0.7566	1.2301	0.0240	0.6615	0.2673	0.0082
EEMD-BP	0.9254	0.6811	0.0113	0.7507	0.2286	0.0073
EEMD-PSO-BP	0.9301	0.6594	0.6594	0.7796	0.2149	0.0058
EEMD-LSSVM	0.9342	0.6240	0.0109	0.8661	0.1675	0.0052
EEMD-GA-ELM	0.9387	0.6175	0.0100	0.8568	0.1739	0.0053
CEEMDAN-ELM	0.8420	0.9901	0.0171	0.8807	0.1581	0.0050
CEEMDAN-BP	0.8916	0.8201	0.0149	0.8813	0.1583	0.0049
CEEMDAN-PSO-BP	0.8969	0.7998	0.0138	0.9157	0.1329	0.0039
CEEMDAN-LSSVM	0.9221	0.6951	0.0116	0.9230	0.1274	0.0039
CEEMDAN-GA-ELM	0.9468	0.5748	0.0099	0.9465	0.1062	0.0032

Download PDF

Version 1

posted

You are reading this latest preprint version

Optimized Extreme Learning Machine (ELM) Based on Genetic Algorithm (GA) To Predict Carbon Price Under The Influence of Multiple Factors

Status:

Version 1

Abstract

Figures

Full Text

Tables

Status:

Version 1