Search results

The purpose of this paper is to find out the best method for forecasting European Union Allowance (EUA) returns and determine its price determinants. The previous studies in this area have focused on a particular subset of EUA data and do not take care of the multicollinearities. The authors take EUA data from all three phases and the continuous series, adopt the principal component analysis (PCA) to eliminate multicollinearities and fit seven different homoscedastic models for a comprehensive analysis.

PCA is adopted to extract independent factors. Seven different linear regression and auto regressive integrated moving average (ARIMA) models are employed for forecasting EUA returns and isolating their price determinants. The seven models are then compared and the one with minimum (root mean square error is adjudged as the best model.

The best model for forecasting the EUA returns of all three phases is dynamic linear regression with lagged predictors and that for forecasting EUA continuous series is ARIMA errors. The latent factors such as switch to gas (STG) and clean spread (capturing the effects of the clean dark spread, clean spark spread, switching price and natural gas price), National Allocation Plan announcements events, energy variables, German Stock Exchange index and extreme temperature events have been isolated as the price determinants of EUA returns.

The current study contributes to effective carbon management by providing a quantitative framework for analyzing cap-and-trade schemes.

This study differs from earlier studies mainly in three aspects. First, instead of focusing on a particular subset of EUA data, it comprehensively analyses the data of all the three phases of EUA along with the EUA continuous series. Second, it expressly adopts PCA to eliminate multicollinearities, thereby reducing the error variance. Finally, it evaluates both linear and non-linear homoscedastic models incorporating lags of predictor variables to isolate the price determinants of EUA.

With the rise of online discussion and argument mining, methods that are able to analyze arguments become increasingly important. A recent study proposed the usage of agreement between arguments to represent both stance polarity and intensity, two important aspects in analyzing arguments. However, this study primarily focused on finetuning bidirectional encoder representations from transformer (BERT) model. The purpose of this paper is to propose convolutional neural network (CNN)-BERT architecture to improve the previous method.

The used CNN-BERT architecture in this paper directly uses the generated hidden representation from BERT. This allows for better use of the pretrained BERT model and makes finetuning the pretrained BERT model optional. The authors then compared the CNN-BERT architecture with the method proposed in the previous study (BERT and Siamese-BERT).

Experiment results demonstrate that the proposed CNN-BERT is able to achieve a 71.87% accuracy in measuring agreement between arguments. Compared to the previous study that achieve an accuracy of 68.58%, the CNN-BERT architecture was able to increase the accuracy by 3.29%. The CNN-BERT architecture is also able to achieve a similar result even without further pretraining the BERT model.

The principal originality of this paper is the proposition of using CNN-BERT to better use the pretrained BERT model for measuring agreement between arguments. The proposed method is able to improve performance and also able to achieve a similar result without further training the BERT model. This allows separation of the BERT model from the CNN classifier, which significantly reduces the model size and allows the usage of the same pretrained BERT model for other problems that also did not need to finetune their BERT model.

The purpose of this study is to predict the thermal protective performance (TPP) of flame-retardant fabric more economically using machine learning and analyze the factors affecting the TPP using model visualization.

A total of 13 machine learning models were trained by collecting 414 datasets of typical flame-retardant fabric from current literature. The optimal performance model was used for feature importance ranking and correlation variable analysis through model visualization.

Five models with better performance were screened, all of which showed R2 greater than 0.96 and root mean squared error less than 3.0. Heat map results revealed that the TPP of fabrics differed significantly under different types of thermal exposure. The effect of fabric weight was more apparent in the flame or low thermal radiation environment. The increase in fabric weight, fabric thickness, air gap width and relative humidity of the air gap improved the TPP of the fabric.

The findings suggested that the visual analysis method of machine learning can intuitively understand the change trend and range of second-degree burn time under the influence of multiple variables. The established models can be used to predict the TPP of fabrics, providing a reference for researchers to carry out relevant research.

The findings of this study contribute directional insights for optimizing the structure of thermal protective clothing, and introduce innovative perspectives and methodologies for advancing heat transfer modeling in thermal protective clothing.

The purpose of this paper is to determine differences in customers’ personal disposition to online shopping.

The research model was proposed based on two types of purchases (hedonic vs utilitarian) and on personal traits of individuals against technology throughout the Technology Readiness Index (TRI) 2.0. Generation and gender were considered to evaluate their impact on the type of purchases. Consumers’ data were collected in Chile through 788 face-to-face surveys. The partial least squares approach was used to test the research model.

The findings show that optimism and discomfort influence online shopping. Moreover, generation and gender moderate the relationship between the dimensions of the TRI and online purchases.

The contributions of this study are threefold. The analysis of personal traits and the type of purchases contribute to the existing literature on consumer behavior and e-commerce, and provide some insights for marketers to identify segmentation strategies by analyzing the gender and generation of individuals. Second, this study contributes to examining the stability and invariances of the TRI 2.0 instrument, which has not been fully revised in less developed countries. Third, this study adds to the existing body of research that argues that demographic variables are not sufficient to understand technology adoption by individuals by including psychological variables.

Predicting the stock market's prices has always been an interesting topic since its closely related to making money. Recently, the advances in natural language processing (NLP) have opened new perspectives for solving this task. The purpose of this paper is to show a state-of-the-art natural language approach to using language in predicting the stock market.

In this paper, the conventional statistical models for time-series prediction are implemented as a benchmark. Then, for methodological comparison, various state-of-the-art natural language models ranging from the baseline convolutional and recurrent neural network models to the most advanced transformer-based models are developed, implemented and tested.

Experimental results show that there is a correlation between the textual information in the news headlines and stock price prediction. The model based on the GRU (gated recurrent unit) cell with one linear layer, which takes pairs of the historical prices and the sentiment score calculated using transformer-based models, achieved the best result.

This study provides an insight into how to use NLP to improve stock price prediction and shows that there is a correlation between news headlines and stock price prediction.

While there is an abundant amount of literature studies on mobile payment adoption, there is a scarce of knowledge concerning the sustainable use of mobile payment contactless technologies. As those technologies are mainly concerned with security and users' trust, the question of how security factors and trust can influence the sustainable use of those technologies within and beyond the COVID-19 pandemic is still unanswered. This research thus develops a theoretical model based on integrating the protection motivation theory (PMT) and the expectation-confirmation model (ECM), extended with perceived trust (PT) to explore the sustainable use of mobile payment contactless technologies.

The developed model is evaluated based on data collected through a web-based survey from 523 users who used contactless payment technologies. Unlike the existing literature, the collected data were analyzed using a hybrid structural equation modeling-artificial neural network (SEM-ANN) technique.

The data analysis results reinforced all the proposed relationships in the developed model. The sensitivity analysis results showed that PT has the largest impact on the sustainable use of mobile payment contactless technologies with 97.2% normalized importance, followed by self-efficacy (SE) (77%), satisfaction (72.1%), perceived vulnerability (PV) (48.9%), perceived usefulness (PU) (48.2%), perceived severity (PS) (40.7%), response efficacy (RE) (28.7%) and response costs (RCs) (24.1%).

The originality of this research lies behind the development of an integrated model based on PMT and ECM to understand the sustainable use of mobile payment contactless technologies. The study provides several managerial implications for decision-makers, policy-makers and service providers to ensure the sustainability of those contactless technologies within and beyond the COVID-19 pandemic.

Abnormal changes in temperature directly affect the stability and reliability of a gyroscope. Predicting the temperature and detecting the abnormal change is great value for timely understanding of the working state of the gyroscope. Considering that the actual collected gyroscope shell temperature data have strong non-linearity and are accompanied by random noise pollution, the prediction accuracy and convergence speed of the traditional method need to be improved. The purpose of this paper is to use a predictive model with strong nonlinear mapping ability to predict the temperature of the gyroscope to improve the prediction accuracy and detect the abnormal change.

In this paper, an double hidden layer long-short term memory (LSTM) is presented to predict temperature data for the gyroscope (including single point and period prediction), and the evaluation index of the prediction effect is also proposed, and the prediction effects of shell temperature data are compared by BP network, support vector machine (SVM) and LSTM network. Using the estimated value detects the abnormal change of the gyroscope.

By combined simulation calculation with the gyroscope measured data, the effect of different network hyperparameters on shell temperature prediction of the gyroscope is analyzed, and the LSTM network can be used to predict the temperature (time series data). By comparing the performance indicators of different prediction methods, the accuracy of the shell temperature estimation by LSTM is better, which can meet the requirements of abnormal change detection. Quick and accurate diagnosis of different types of gyroscope faults (steps and drifts) can be achieved by setting reasonable data window lengths and thresholds.

The LSTM model is a deep neural network model with multiple non-linear mapping levels, and can abstract the input signal layer by layer and extract features to discover deeper underlying laws. The improved method has been used to solve the problem of strong non-linearity and random noise pollution in time series, and the estimated value can detect the abnormal change of the gyroscope.

In this paper, based on the LSTM network, an double hidden layer LSTM is presented to predict temperature data for the gyroscope (including single point and period prediction), and validate the effectiveness and feasibility of the algorithm by using shell temperature measurement data. The prediction effects of shell temperature data are compared by BP network, SVM and LSTM network. The LSTM network has the best prediction effect, and is used to predict the temperature of the gyroscope to improve the prediction accuracy and detect the abnormal change.

This study explores the adoption of Industry 4.0 in developing countries' export industries, focusing on factors influencing this adoption, the moderating role of market pressure and prioritizing key factors for sustainable growth.

Based on the “TOE theory” this study has proposed a research framework to identify the factors influencing the adoption and sustainable implementation of Industry 4.0 in the export industry. This study has collected valid datasets from 387 export-oriented industries and applied SEM-ANN dual-stage hybrid model to capture linear and nonlinear interaction between variables.

Results revealed that Technical Capabilities, System Flexibility, Software Infrastructure, Human Resource Competency and Market pressure significantly influence the Adoption of Industry 4.0. Higher market pressure as a moderator also improves the Industry 4.0 adoption process. Results also pointed out that system flexibility is a gray area in Industry 4.0 adoption, which can be enhanced in the export industry to maintain a sustainable adoption and implementation of Industry 4.0.

Minute information is available on the factors influencing the adoption of Industry 4.0 in export-oriented industries. This study has empirically explored the role of influential factors in Industry 4.0 and ranked them based on their normalized importance.

In this study the authors aim to outline new ways of information extraction for automated valuation models, which in turn would help to increase transparency in valuation procedures and thus contribute to more reliable statements about the value of real estate.

The authors hypothesize that empirical error in the interpretation and qualitative assessment of visual content can be minimized by collating the assessments of multiple individuals and through use of repeated trials. Motivated by this problem, the authors developed an experimental approach for semi-automatic extraction of qualitative real estate metadata based on Comparative Judgments and Deep Learning. The authors evaluate the feasibility of our approach with the help of Hedonic Models.

The results show that the collated assessments of qualitative features of interior images show a notable effect on the price models and thus over potential for further research within this paradigm.

To the best of the authors’ knowledge, this is the first approach that combines and collates the subjective ratings of visual features and deep learning for real estate use cases.

This paper aims to deal with the optimal choice of a novel extreme learning machine (ELM) architecture based on an ensemble of classic ELM called Meta-ELM structural parameters by using a forecasting process.

The modelling performance of the Meta-ELM architecture varies depending on the network parameters it contains. The choice of Meta-ELM parameters is important for the accuracy of the models. For this reason, the optimal choice of Meta-ELM parameters is investigated on the problem of wind speed forecasting in this paper. The hourly wind-speed data obtained from Bilecik and Bozcaada stations in Turkey are used. The different number of ELM groups (M) and nodes (N_h) are analysed for determining the best modelling performance of Meta-ELM. Also, the optimal Meta-ELM architecture forecasting results are compared with four different learning algorithms and a hybrid meta-heuristic approach. Finally, the linear model based on correlation between the parameters was given as three dimensions (3D) and calculated.

It is observed that the analysis has better performance for parameters of Meta-ELM, M = 15 − 20 and N_h = 5 − 10. Also considering the performance metric, the Meta-ELM model provides the best results in all regions and the Levenberg–Marquardt algorithm -feed forward neural network and adaptive neuro fuzzy inference system -particle swarm optimization show competitive results for forecasting process. In addition, the Meta-ELM provides much better results in terms of elapsed time.

The original contribution of the study is to investigate of determination Meta-ELM parameters based on forecasting process.