Search

Financial time series forecasting is crucial in empowering investors to make well-informed decisions, manage risks effectively, and strategically plan their investment activities. However, the non-stationary and non-linear characteristics inherent in time series data pose significant challenges when accurately predicting future forecasts. This paper proposes a novel Recurrent ensemble deep Random Vector Functional Link (RedRVFL) network for financial time series forecasting. The proposed model leverages randomly initialized and fixed weights for the recurrent hidden layers, ensuring stability during training. Furthermore, incorporating stacked hidden layers enables deep representation learning, facilitating the extraction of complex patterns from the data. The proposed model generates the forecast by combining the outputs of each layer through an ensemble approach. A comparative analysis was conducted against several state-of-the-art models over financial time-series datasets, and the results demonstrated the superior performance of our proposed model in terms of forecasting accuracy and predictive capability. [ABSTRACT FROM AUTHOR]

Neural networks have been successfully employed in various domains such as classification, regression and clustering, etc. Generally, the back propagation (BP) based iterative approaches are used to train the neural networks, however, it results in the issues of local minima, sensitivity to learning rate and slow convergence. To overcome these issues, randomization based neural networks such as random vector functional link (RVFL) network have been proposed. RVFL model has several characteristics such as fast training speed, direct links, simple architecture, and universal approximation capability, that make it a viable randomized neural network. This article presents the first comprehensive review of the evolution of RVFL model, which can serve as the extensive summary for the beginners as well as practitioners. We discuss the shallow RVFLs, ensemble RVFLs, deep RVFLs and ensemble deep RVFL models. The variations, improvements and applications of RVFL models are discussed in detail. Moreover, we discuss the different hyperparameter optimization techniques followed in the literature to improve the generalization performance of the RVFL model. Finally, we present potential future research directions/opportunities that can inspire the researchers to improve the RVFL's architecture and learning algorithm further. • The first survey focusing solely on RVFL-based models. • RVFLs in shallow, ensemble, deep, and ensemble deep frameworks have been discussed. • Various applications of the RVFL have been discussed. • Hyper-parameter optimization and experimental setups for the RVFL are discussed. • We present potential future research directions for the RVFL model. [ABSTRACT FROM AUTHOR]

Accurate forecasting of the air quality index (AQI) plays a crucial role in taking precautions against upcoming air pollution risks. However, air quality may fluctuate greatly in a certain period. Existing forecasting approaches always face the problem of losing valuable information on air quality status, even in the interval models of recent research. To address this issue, this paper suggests a new AQI forecasting perspective and paradigm built upon ternary interval-valued time series (TITS), multivariate variational mode decomposition (MVMD), multivariate relevance vector machine (MVRVM), mixed coding particle swarm optimization (MCPSO), and meteorological factors, which is able to capture the trend and volatility changes of AQI concurrently. The proposed paradigm involves four procedures: TITS construction in terms of the daily minimum, daily mean, and daily maximum AQI, multi-scale decomposition via MVMD, individual forecasting by MCPSO-optimized MVRVM, and ensemble learning forecasting using a simple addition approach. Experiments based on datasets collected from four municipalities in China demonstrated that the presented paradigm can hit higher accuracy than other comparable models, and the application analysis also shows that it has application potential in the AQI online forecasting system. To conclude, the proposed paradigm provides a promising alternative for AQI time series forecasting. • A ternary interval decomposition ensemble learning paradigm for AQI time series forecasting. • Multivariate variational mode decomposition is applied to thoroughly analyze the TIAQITS. • Multivariate relevance vector machine is introduced to model the three bounds of TIAQITS. • Mixed coding particle swarm optimization is adopted to optimize the parameters of MVRVM. [ABSTRACT FROM AUTHOR]

Traffic forecasting is crucial to achieving a smart city as it facilitates public transportation management, autonomous driving, and the resource relocation of the sharing economy. Traffic forecasting belongs to the challenging spatiotemporal forecasting task, which is highly demanding because of the complicated geospatial correlation between traffic nodes, inconsistent and highly non-linear temporal patterns due to various events, and sporadic traffic accidents. Previous graph neural network (GNN) models built for transportation forecasting feature the sophisticated structure and heavy computation cost as they combine the deep neural network and graph machine learning to capture the spatiotemporal dynamics for the whole transportation network. However, it may be more practical for practitioners to perform node-wise forecasting for specific nodes of interest rather than network-wise forecasting. To mitigate the gaps mentioned above, we propose a novel graph ensemble deep random vector functional link network (GEdRVFL) to forecast the future traffic volume by combining the well-performing ensemble deep random vector functional link (EdRVFL) with the graph convolution layer for a specific node and realize the node-wise traffic forecasting. After a comprehensive comparison with the state-of-the-art models, our model beats the others in four out of five cases measured by mean absolute scaled error. • A graph EdRVFL network for node-wise spatiotemporal forecasting is proposed. • The intermediate features are filtered through a model-based selection method. • The search of hyperparameters is implemented using Bayesian optimization algorithm. • Our method outperforms the state-of-the-art models on a large-scale dataset. [ABSTRACT FROM AUTHOR]

Time charter rates must be predicted accurately to assist sensible decisions in the global, highly volatile shipping market. Time charter rates are affected by multiple factors, such as second-hand ship prices, order book, Libor interest rate, etc. However, not all factors convey predictive features to anticipate the future of time charter rates. Therefore, extracting predictive features from multiple driving time series from the shipping market is crucial for forecasting purposes. Accordingly, this paper proposes a novel convolutional recurrent neural network for time charter rates forecasting under the multi-variate phenomenon. The proposed network first extracts features from the monthly time series using a novel convolutional filter, the annual dilated filter. The annual dilated convolutional filter can extract the predictive features effectively and impose a sparse input structure to prevent overfitting. Then, a recurrent neural network learns the temporal information from the convoluted features. An extensive comparison study with many baseline models, including the persistence (Naïve I), statistical models, and the state-of-art networks, is conducted on the time charter rates of six kinds of ships. The empirical results demonstrate the proposed model's superiority in forecasting the time charter rates. • A novel dilated convolution LSTM neural network is proposed to forecast time charter rates. • The proposed convolutional filters can extract temporal features of monthly charter rates efficiently. • The proposed model outperforms the baselines in terms of three error metrics. [ABSTRACT FROM AUTHOR]

The implementation of International Maritime Organization (IMO) 2020 sulfur cap requires ship operators to decrease fuels' sulfur content and this may increase their demand for low-sulfur fuel (LSF). In anticipation, bunker companies can choose to upgrade their refining technology to produce better quality distillate and lighter oil. In this study, we consider a bunker supply chain consisting of bunker companies and a population of ship operators with two main marine fuel products, low-sulfur fuel (LSF) and high-sulfur fuel (HSF). We use Cournot game to model the competition between LSF and HSF under two different market channels (i.e., dual and single channels). The results show that bunker companies' refining technology upgrading choice is affected by many operational parameters, such as the basic market demand, cost difference between LSF and HSF, market competition, variable and fixed cost for upgrading, the increase of market demand due to upgrading, and so on. Compared with a dual channel bunker company, a single channel bunker company is less likely to implement new refining technology. We further consider the scenario where bunker companies can make decisions after the realization of ship operators' demand uncertainty. The findings are beneficial for both LSF and HSF bunker companies and may reach a win-win solution for bunker companies. [ABSTRACT FROM AUTHOR]

The challenge of accurately forecasting a time series covers numerous disciplines, from economics to engineering. Among the thousands of machine learning models, random vector functional link (RVFL) is a robust and efficient model which has demonstrated its success in various challenging forecasting problems. RVFL is an efficient universal function appropriator that randomly generates the weights between the input and hidden layers. However, RVFL still lacks the strong ability to extract meaningful multi-scale features from input data because of the single-layer random mapping of enhancement nodes. Therefore, we propose to combine the empirical wavelet transformation (EWT) with RVFL to strengthen the multi-scale feature extraction ability. The EWT can decompose the original time series into several sub-series which carry the information of different frequencies. Besides, we propose a walk-forward decomposition mechanism to implement the EWT. By introducing such a walk-forward mechanism and the combination of EWT and RVFL, the hybrid model achieves high accuracy and averts the data leakage problem during forecasting. A detailed and comprehensive empirical study on twenty-six public time series validates the proposed model's superiority compared with ten popular baseline models from the literature. • A causal signal decomposition method is proposed to avert the data leakage problem. • The empirical wavelet transformation is applied to decompose the time series. • A novel hybrid RVFL network is proposed for time series forecasting. [ABSTRACT FROM AUTHOR]

Echo state network (ESN) is a reservoir computing structure consisting randomly generated hidden layer which enables a rapid learning and extrapolation process. On the other hand, the determination of inputs is still an outstanding question similar to other neural networks. Instead of utilizing the original historical data, wavelet transformations can be used to extract and reflect features with predictability (periodicity) and eliminate random oscillations (i.e. noise) which would distort the underlying predictive structure. In this regard, we propose a two-stage predictive algorithm in which the empirical wavelet transformation (EWT) has been implemented to transform the data, and ESN is utilized to execute the overall predictive process. The proposed method has been validated by twelve public datasets with different mean-volatility features. Out of sample forecasts have been compared to the baseline persistence model and conventional benchmarks. The empirical study demonstrates the superiority of the EWT–ESN model compared with other models. • A novel EWT–ESN is proposed for time series forecasting. • The EWT is developed to decompose the original time series. • A rapid predictive process has been developed with ESN structure. • Twelve datasets are used to validate the superiority of the proposed model. [ABSTRACT FROM AUTHOR]