Soft computing prediction of economic growth based in science and technology factors

doi:10.1016/j.physa.2016.08.034

Physica A: Statistical Mechanics and its Applications

Volume 465, 1 January 2017, Pages 217-220

https://doi.org/10.1016/j.physa.2016.08.034 Get rights and content

Highlights

•
The estimation of the gross domestic product (GDP) growth rate.
•
Economic growth basis on combination of different factors.
•
The accuracy of the extreme learning machine (ELM).

Abstract

The purpose of this research is to develop and apply the Extreme Learning Machine (ELM) to forecast the gross domestic product (GDP) growth rate. In this study the GDP growth was analyzed based on ten science and technology factors. These factors were: research and development (R&D) expenditure in GDP, scientific and technical journal articles, patent applications for nonresidents, patent applications for residents, trademark applications for nonresidents, trademark applications for residents, total trademark applications, researchers in R&D, technicians in R&D and high-technology exports. The ELM results were compared with genetic programming (GP), artificial neural network (ANN) and fuzzy logic results. Based upon simulation results, it is demonstrated that ELM has better forecasting capability for the GDP growth rate.

Introduction

Gross Domestic Product (GDP) is the main feature for economic growth. GDP prediction could be challenging task because of uncertain factors. More advances in prediction technique for GDP are needed. In order to reduce wasting many decisions want to know in advance what will be system behavior in economic aspect. In other words investors require to know future technology success. There were several approaches so far for the GDP modeling and forecasting [1]. More complexity has positive impact on economic growth [2]. Article [3] was shown that the artificial neural network (ANN) can produce better results for economic growth forecasting than other methods. There is positive correlation between the forecasters’ number and the forecast accuracy [4]. GDP prediction accuracy can improve by expansion in information domain [5], [6]. The growth of GDP was considered as a process to describe the logistic-growth equation [7].

Decisions in any technology investments are influenced by different factors such as human resources, finance investment etc. Prediction of the investment benefits in new technologies is very important in order to reduce resource wasting. The success of the new technologies forecasting is crucial for decision makers. The investors require to know future technology success in order to make investments decision. Patent applications can be used as indicator for prediction of success of technology. Patents and innovations can influence GDP growth rate also.

Even though a number of new mathematical functions have been proposed for modeling of the GDP growth analysis [8], [9], [10], in this investigation extreme learning machine (ELM) [11], [12] was used for GDP growth rate prediction based on science and technology factors The objective was to analyze the economic growth forecasting based on the research and development (R&D) expenditure in GDP, scientific and technical journal articles, patent applications for nonresidents, patent applications for residents, trademark applications for nonresidents, trademark applications for residents, total trademark applications, researchers in R&D, technicians in R&D and high-technology exports. ELM is capable to solve problems caused by gradient descent based algorithms and to decrease required time for training. ELM has been widely applied for the estimation in many different fields so far [13], [14], [15], [16].

Section snippets

Database

In this investigation ten science and technology parameters were used as the inputs for the GDP prediction. The data were from 28 countries in the European Union. Table 1 shows all input parameters which were used in this study.

Extreme learning machine

Extreme Learning Machine (ELM) was developed as a learning algorithm for artificial neural networks (ANN) [13], [14], [15], [16]. This approach has advantages as compared with conventional learning algorithms. The main advantage of the ELM methodology is fast training

Results

Fig. 2 shows GDP prediction accuracy results with ELM methodology and also other soft computing methods as comparison. Elm method outperforms ANN [17] and GP [18], fuzzy [19] methods according to the coefficient of determination. In order to demonstrate the merits of the proposed model on a more tangible basis, four models’ prediction accuracy was compared each other. Standard statistical indicators, RMSE, $r$ and $R^{2}$ were used for comparison. Table 2 summarizes the prediction accuracy results

Conclusion

In the present study, we describe the prediction of GDP prediction using the ELM method was presented in this article that showcases its benefits to predict the GDP growth rate. In this study science and technology factors are used as inputs for GDP growth rate. GDP growth rate prediction is very difficult task due to many indicators and factors which influence the GDP.

A learning model based on ELM was developed to predict the GDP growth rate. Accuracy level of predicted values was assessed in

References (19)

Joseph Zeira et al.
Economic growth and sector dynamics
Eur. Econ. Rev.
(2015)
Benno Ferrarini et al.
Production complexity, adaptability and economic growth
Struct. Change Econ. Dyn.
(2016)
Lihua Feng et al.
Application of artificial neural networks in tendency forecasting of economic growth
Econ. Modell.
(2014)
Libor Krkoska et al.
How reliable are forecasts of GDP growth and inflation for countries with limited coverage?
Econ. Syst.
(2009)
Libor Krkoska et al.
Accuracy of GDP growth forecasts for transition countries: Ten years of forecasting assessed
Int. J. Forecast.
(2007)
Francisco Dias et al.
Forecasting Portuguese GDP with factor models: Pre- and post-crisis evidence
Econ. Modell.
(2015)
Theodore Modis
Long-term GDP forecasts and the prospects for growth
Technol. Forecast. Soc. Change
(2013)
Fady Barsoum et al.
Forecasting GDP growth using mixed-frequency models with switching regimes
Int. J. Forecast.
(2015)
R. Nian et al.
Extreme learning machine towards dynamic model hypothesis in fish ethology research
Neurocomputing
(2014)

There are more references available in the full text version of this article.

Cited by (24)

Optimized deep extreme learning machine for traffic prediction and autonomous vehicle lane change decision-making
2024, Physica A: Statistical Mechanics and its Applications
Deep learning occupies a prominent position in the field of machine learning, where researchers have introduced autoencoder into extreme learning machine (ELM) to establish deep extreme learning machine (DELM), with the aim of enabling autonomous feature learning from data. This approach has found widespread applications in various domains, including finance and electrochemistry. However, the network accuracy of DELM is sensitive to the initial weighting and bias configurations, which can be negatively impacted by random initialization. To mitigate the adverse effects of such random initialization on network performance, this study employs two swarm intelligence optimization algorithms, particle swarm optimization (PSO) and harris hawks optimization (HHO), to optimize the input weights and biases of DELM. In addition, we apply the variational mode decomposition (VMD) technique to decompose the original data, enabling a deeper exploration of latent information within the data. Subsequently, two integrated models are constructed: VMD-HHO-DELM and VMD-PSO-DELM. We apply these models to address road traffic condition prediction and autonomous lane change decision-making problems, with a specific focus on the transportation domain. Through extensive experimentation on two real-world datasets, we observe that deep learning models based on swarm intelligence optimization algorithms exhibit excellent performance across various scenarios. In the context of road traffic condition prediction, our experiments reveal superior results under different time intervals and traffic volumes compared to traditional models, all while maintaining higher operational efficiency. For the autonomous lane change decision experiment, we establish a driver dissatisfaction model that aids in formulating more reasonable lane change decisions, thus reducing the frequency of abrupt lane changes.
Prediction and optimization of the flexural behavior of corroded concrete beams using adaptive neuro fuzzy inference system
2022, Structures
Citation Excerpt :
The last layers added together all of the signals while providing the output crisp value. The current NN was built by training and testing through empirical data from the literature [14,42–46] for 107 corroded concrete components. Table 1 shows the range of data collected throughout the test inquiry.
Global analysis on the number of faulty bridges, together with continuing corrosion procedure is kept on by deicing chemicals in various climates that create a necessity toward improved analytical procedures for reinforced concrete elements damaged by corrosion. Soft computing approaches could be used to simulate these statues i.e., finite element (FE) is a perfect tool for meeting this demand. Nonetheless, these assessments need a large number of inputs that, due to the extended periods of corrosion occurring, are sometimes too expensive to gather via physical testing. Here, a new statistical method as adaptive neuro fuzzy inference system (ANFIS) including data from 107 concrete members was developed to estimate these inputs. Regression models are created and analyzed which is the main novelty of the work. The resulting graphs from such ANFIS models demonstrate strong correlation that supporting the ANFIS's precision. As a result, the ANFIS is proposed as a method to define the flexural behavior of concrete members damaged by corrosion.
Decoupling analysis of economic growth and mineral resources consumption in China from 1992 to 2017: A comparison between tonnage and exergy perspective
2021, Resources Policy
Citation Excerpt :
The former will increase the decoupling index, while the latter will make the decoupling index fall. Therefore, in the GRA model, the GDP growth rate (x7), the share of exports of goods and services in GDP (x8) and the share of net inflows of foreign direct investment in GDP (x9) are used to represent the growth pattern of China's economy (Fotis et al., 2017; Marković et al., 2017). It should be pointed out that compared with previous studies (e.g., Liu et al., 2016; Ma et al., 2017; Song et al., 2019), the types of mineral resources involved in this study are more extensive.
The utilization of mineral resources not only acts as one of the fundamental aspects supporting economic activities, but also leads to resource depletion and ecological environment problems. To achieve economic growth without increases in quantitative mineral resources consumption and adverse environmental effects, it is necessary to conduct a decoupling analysis combining resource decoupling and environmental impact decoupling simultaneously. This study introduces the exergoecological approach into the decoupling analysis framework to investigate the decoupling of China's economic growth from mineral resources consumption and its environmental burden. On this basis, the grey relation analysis (GRA) is adopted to explore influence factors of the decoupling index from three aspects (the technological progress, the industrial structure and the economic growth mode). The results show that: (1) the environmental burden generated by the consumption of building minerals is equivalent to the sum of non-ferrous metal minerals and salt minerals, and almost all the environmental burden of non-fuel minerals consumption comes from the above three categories; (2) the quality of mineral resources has a significant impact on the decoupling index, especially for non-fuel minerals, and the decoupling index measured in exergy replacement costs (ERC) is smaller than that calculated in tons; (3) the decoupling index of fossil fuels is most influenced by the industrial structure, followed by the technological progress and the economic growth mode, while the decoupling index of non-fuel minerals is more influenced by the technological progress and the industrial structure than the economic growth mode. Based on the research results, this study can serve as a foundation to offer decision makers relevant implications for formulation of sustainable development policies.
Novel mixed-encoding for forecasting patent grant duration
2021, World Patent Information
Citation Excerpt :
By applying computational knowledge techniques, financial improvement estimation can be decided dependent on various science and innovation factors [12]. GDP growth rate was forecasted using extreme machine learning (ELM) by Markovic et al. [13]. Lee et al. [14] propose a quantitative performance prediction for corporate sector model that applies the Support Vector Regression algorithm from technical and financial data.
In an age when data is regarded as the most essential commodity, organizations are racing to use it for better decision making. The quality of the patent portfolio is an important indicator of technological innovation in an organization and its analysis can reveal several indicators linked to the growth of a company. The advancement of machine learning along with the access to large amounts of patent data has led to a paradigm shift from traditional patent data analysis methodologies to novel approaches. A lot of research has been done in this direction for analysing data on patent citations, patent text, IPC class etc. However, much less has been explored regarding the forecast of patent grant duration and its significance for decision making with an even lower focus on data collected from developing countries. This work is built upon our existing study on patent grant duration prediction by devising a novel methodology of encoding the data using a combination of augmented one-hot encoding and label-encoding. Thereafter, methodologies such as Outlier Detection have been applied to this data to yield an improved result vis-à-vis our baseline results. In addition, we identify some of the important factors which impact the decision on grant duration of patent applications using the raw data from the Indian Patent Office.
Prediction of the flexural behavior of corroded concrete beams using combined method
2020, Structures
Citation Excerpt :
The present endeavor develops an analytical method combining neural network and finite element in ABAQUS software to predict the performance of corroded reinforced concrete beams. Neural network (NN) is regarded as a form of computational intelligence to imitate the human brain and nervous system in a simple manner [20,27–30]. Substantially, the structure of neural networks is composed of processing elements with highly interconnection [21].
Nationwide statistics on numbers of structurally-deficient bridges coupled with ongoing corrosion processes caused by deicing agents in many climates lead to a demand for better analysis techniques for corrosion-damaged reinforced concrete structural members. Modern computational methods for modeling this behavior such as finite element analysis (FEA) are an ideal tool to fulfill this need. However, these analyses require many inputs that, due to the long timescales over which corrosion occurs, are often prohibitive to obtain through physical testing. In this research, a novel statistical approach using a neural network (NN) model has been constructed to approximate these inputs based on data in the literature from 107 concrete members. Then, outputs from the NN have been introduced into FEA material behavior models for the analysis of corrosion-damaged concrete beams. Load-deflection behavior resulting from such FEA shows good correlation when compared with available experimental data, confirming the accuracy of the NN. Thus, the NN is suggested as a means for obtaining inputs for FEA of corrosion-damaged concrete members.
Unemployment expectations: A socio-demographic analysis of the effect of news
2019, Labour Economics
Citation Excerpt :
Yang et al. (2015) used a data-driven approach based on SR to predict oil production in the US, using data from the 48 lower states since 1859. Marković et al. (2017) made use of 10 science and technology factors as inputs for GDP growth prediction in 28 EU countries. More recently, Claveria et al. (2018) applied a GP procedure to derive a set of building blocks used to estimate economic activity in the main European countries, finding that the proposed economic indicator outperformed a random walk model used as a benchmark in most cases.
In this study, we evaluate the effect of news on consumer unemployment expectations for sixteen socio-demographic groups. To this end, we construct an unemployment sentiment indicator and extract news about several economic variables. By means of genetic programming we estimate symbolic regressions that link unemployment rates in the Euro Area to qualitative expectations about a wide range of economic variables. We then use the evolved expressions to compute unemployment expectations for each consumer group. We first assess the out-of-sample forecast accuracy of the evolved indicators, obtaining better forecasts for the leading unemployment sentiment indicator than for the coincident one. Results are similar across the different socio-demographic groups. The best forecast results are obtained for respondents between 30 and 49 years. The group where we observe the bigger differences among categories is the occupation, where the lowest forecast errors are obtained for the unemployed respondents. Next, we link news about inflation, industrial production, and stock markets to unemployment expectations. With this aim we match positive and negative news with consumers’ unemployment sentiment using a distributed lag regression model for each news item. We find asymmetries in the responses of consumers’ unemployment expectations to economic news: they tend to be stronger in the case of negative news, especially in the case of inflation.

View all citing articles on Scopus

View full text

Soft computing prediction of economic growth based in science and technology factors

Highlights

Abstract

Introduction

Section snippets

Database

Extreme learning machine

Results

Conclusion

Eur. Econ. Rev.

Struct. Change Econ. Dyn.

Econ. Modell.

Econ. Syst.

Int. J. Forecast.

Econ. Modell.

Technol. Forecast. Soc. Change

Int. J. Forecast.

Neurocomputing