Forecasting

Accurate and efficient cryptocurrency price prediction is vital for investors in the volatile crypto market. This study comprehensively evaluates nine models—including baseline, zero-shot, and deep learning architectures—on 21 major cryptocurrencies using daily and hourly data. Our multi-dimensional evaluation assesses models based on prediction accuracy (MAE, RMSE, MAPE), speed, statistical significance (Diebold–Mariano test), and economic value (Sharpe Ratio). Our research found that the optimally fine-tuned TimeGPT model (without variables) demonstrated superior performance across both Daily and Hourly datasets, with its statistical leadership confirmed by the Diebold–Mariano test. Fine-tuned Chronos excelled in daily predictions, while TFT was a close second to TimeGPT for hourly forecasts. Crucially, zero-shot models like TimeGPT and Chronos were tens of times faster than traditional deep learning models, offering high accuracy with superior computational efficiency. A key finding from our economic analysis is that a model’s effectiveness is highly dependent on market characteristics. For instance, TimeGPT with variables showed exceptional profitability in the volatile ETH market, whereas the zero-shot Chronos model was the top performer for the cyclical BTC market. This also highlights that variables have asset-specific effects with TimeGPT: improving predictions for ICP, LTC, OP, and DOT, but hindering UNI, ATOM, BCH, and ARB. Recognizing that prior research has overemphasized prediction accuracy, this study provides a more holistic and practical standard for model evaluation by integrating speed, statistical significance, and economic value. Our findings collectively underscore TimeGPT’s immense potential as a leading solution for cryptocurrency forecasting, offering a top-tier balance of accuracy and efficiency. This multi-dimensional approach provides critical, theoretical, and practical guidance for investment decisions and risk management, proving especially valuable in real-time trading scenarios. Full article

This study presents a gradient-informed proxy initialization framework designed to improve training efficiency and predictive performance in deep learning models for time-series forecasting. The method extends the Laor Initialization approach by introducing backward gradient norm clustering as a selection criterion for input-layer weights, evaluated through a lightweight, architecture-agnostic proxy model. Only the numerical input layer adopts the selected initialization, while internal components retain standard schemes such as Xavier, Kaiming, or Orthogonal, maintaining compatibility and reducing overhead. The framework is evaluated on a real-world financial forecasting task: identifying high-risk equities from the Thai Market Surveillance Measure List, a domain characterized by label imbalance, non-stationarity, and limited data volume. Experiments across five architectures, including Transformer, ConvTran, and MMAGRU-FCN, show that the proposed strategy improves convergence speed and classification accuracy, particularly in deeper and hybrid models. Results in recurrent-based models are competitive but less pronounced. These findings support the method’s practical utility and generalizability for forecasting tasks under real-world constraints. Full article

The dry-bulb temperature is a critical parameter in weather forecasting, agriculture, energy management, and climate research. This work proposes a new hybrid prediction model (FBSE-GA-LSTM) that integrates the Fourier–Bessel series expansion (FBSE), genetic algorithm (GA), and long short-term memory (LSTM) networks together to predict the dry-bulb air temperature. The hybrid model FBSE-GA-LSTM utilises the FBSE to decompose time series data of interest into an attempt to remove the noise level for capturing the dominant predictive patterns. Then, the FBSE is embedded into the GA method for the best feature selection and dimension reduction. To predict the dry-bulb temperature, a new model (FBSE-GA-LSTM) was used by hybridising a proposed model FBSE-GA with the LSTM model on the time series dataset of two different regions in Saudi Arabia. For comparison, the FBSE and GA models were hybridised with a bidirectional LSTM (BiLSTM), gated recurrent unit (GRU), and bidirectional gated recurrent unit (BiGRU) models to obtain the hybrid FBSE-GA-BiLSTM, FBSE-GA-GRU, and FBSE-GA-BiGRU models along with their standalone versions. In addition, benchmark models, including the climatic average and persistence approaches, were employed to demonstrate that the proposed model outperforms simple baseline predictors. The experimental results indicated that the proposed hybrid FBSE-GA-LSTM model achieved improved prediction performance compared with the contrastive models for the Jazan region, with a mean absolute error (MAE) of 1.458 °C, a correlation coefficient (R) of 0.954, and a root mean squared error (RMSE) of 1.780 °C, and for the Jeddah region, with an MAE of 1.459 °C, an R of 0.952, and an RMSE of 1.782 °C, between the predicted and observed values of dry-bulb air temperature. Full article

The accurate short-term forecasting (PV) of power is crucial for grid stability control, energy trading optimization, and renewable energy integration in smart grids. However, PV generation is extremely variable and non-linear due to environmental fluctuations, which challenge the conventional forecasting models. This study proposes a hybrid deep learning architecture, Wavelet Transform–Transformer–Temporal Convolutional Network–Efficient Channel Attention Network–Gated Recurrent Unit (WT–Transformer–TCN–ECANet–GRU), to capture the overall temporal complexity of PV data through integrating signal decomposition, global attention, local convolutional features, and temporal memory. The model begins by employing the Wavelet Transform (WT) to decompose the raw PV time series into multi-frequency components, thereby enhancing feature extraction and denoising. Long-term temporal dependencies are captured in a Transformer encoder, and a Temporal Convolutional Network (TCN) detects local features. Features are then adaptively recalibrated by an Efficient Channel Attention (ECANet) module and passed to a Gated Recurrent Unit (GRU) for sequence modeling. Multiscale learning, attention-driven robust filtering, and efficient encoding of temporality are enabled with the modular pipeline. We validate the model on a real-world, high-resolution dataset of a Moroccan university building comprising 95,885 five-min PV generation records. The model yielded the lowest error metrics among benchmark architectures with an MAE of 209.36, RMSE of 616.53, and an R² of 0.96884, outperforming LSTM, GRU, CNN-LSTM, and other hybrid deep learning models. These results suggest improved predictive accuracy and potential applicability for real-time grid operation integration, supporting applications such as energy dispatching, reserve management, and short-term load balancing. Full article

This paper introduces Neighborhood-Constrained Decomposition-based Prediction (NCD-Pred), the first system to leverage Neighborhood-Constrained Variational Mode Decomposition (NCVMD) for multichannel forecasting by integrating time series decomposition and neural networks. NCD-Pred leverages NCVMD to decompose a multichannel signal into simpler, band-limited components—referred to as intrinsic mode functions or simply modes—by prioritizing the most informative channel (the main channel) over less informative ones (the auxiliary channels) and bringing their central frequencies into alignment up to a tunable extent. This frequency synchronization provides a framework for cooperative mode forecasting, where predictions of signal components are recombined to produce the original signal prediction. For mode-level forecasting, Long Short-Term Memory (LSTM) networks are utilized. NCD-Pred’s performance is evaluated against similarly designed mode-level forecasting systems using a multichannel dataset with weak cross-correlation, representing power load on a large vessel. The results show that NCD-Pred outperforms benchmark methods, demonstrating its practical utility in real signal processing scenarios. Full article

The neural architecture search technique is used to automate the engineering of neural network models. Several studies have applied this approach, mainly in the fields of image processing and natural language processing. Its application generally requires very long computing times before converging on the optimal architecture. This study proposes a hybrid approach that combines transfer learning and dynamic search space adaptation (TL-DSS) to reduce the architecture search time. To validate this approach, Long Short-Term Memory (LSTM) models were designed using different evolutionary algorithms, including artificial bee colony (ABC), genetic algorithm (GA), differential evolution (DE), and particle swarm optimization (PSO), which were developed to predict trends in global horizontal irradiation data. The performance measures of this approach include the performance of the proposed models, as evaluated via RMSE over a 24-h prediction window of the solar irradiance data trend on one hand, and CPU search time on the other. The results show that, in addition to reducing the search time by up to 89.09% depending on the search algorithm, the proposed approach enables the creation of models that are up to 99% more accurate than the non-enhanced approach. This study demonstrates that it is possible to reduce the search time of a neural architecture while ensuring that models achieve good performance. Full article

The growing demand for efficient energy management has become essential for achieving sustainable development across social, economic, and environmental sectors. Accurate energy demand forecasting plays a pivotal role in energy management. However, energy demand data present unique challenges due to their complex characteristics, such as multi-seasonality, hidden structures, long-range dependency, irregularities, volatilities, and nonlinear patterns, making energy demand forecasting challenging. We propose a hybrid dimension reduction deep learning algorithm, Temporal Variational Residual Network (TVRN), to address these challenges and enhance forecasting performance. This model integrates variational autoencoders (VAEs), Residual Neural Networks (ResNets), and Bidirectional Long Short-Term Memory (BiLSTM) networks. TVRN employs VAEs for dimensionality reduction and noise filtering, ResNets to capture local, mid-level, and global features while tackling gradient vanishing issues in deeper networks, and BiLSTM to leverage past and future contexts for dynamic and accurate predictions. The performance of the proposed model is evaluated using energy consumption data, showing a significant improvement over traditional deep learning and hybrid models. For hourly forecasting, TVRN reduces root mean square error and mean absolute error, ranging from 19% to 86% compared to other models. Similarly, for daily energy consumption forecasting, this method outperforms existing models with an improvement in root mean square error and mean absolute error ranging from 30% to 95%. The proposed model significantly enhances the accuracy of energy demand forecasting by effectively addressing the complexities of multi-seasonality, hidden structures, and nonlinearity. Full article

Multivariate Time Series Forecasting (MTSF) has been innovated with a series of models in the last two decades, ranging from traditional statistical approaches to RNN-based models. However, recent contributions from deep learning to time series problems have made huge progress with a series of Transformer-based models. Despite the breakthroughs by attention mechanisms applied to deep learning areas, many challenges remain to be solved with more sophisticated models. Existing Transformers known as attention-based models outperform classical models with abilities to capture temporal dependencies and better strategies for learning dependencies among variables as well as in the time domain in an efficient manner. Aiming to solve those issues, we propose a novel Transformer—SegmentedCrossformer (SCF), a Transformer-based model that considers both time and dependencies among variables in an efficient manner. The model is built upon the encoder–decoder architecture in different scales and compared with the previous state of the art. Experimental results on different datasets show the effectiveness of SCF with unique advantages and efficiency. Full article

This study adapts the United Nations’ methodology for national probabilistic population projections to subnational contexts. The Bayesian approach used by the UN addresses data collection complexities effectively. By applying hierarchical model assumptions, national projections can be extended to subnational levels. There is a significant demand for subnational projections with uncertainty measures, especially in South Korea, where low fertility rates have led to rapid population decline, impacting economic and social conditions. The Bayesian hierarchical model predicts South Korea’s population will peak in 2024 and then decline sharply, potentially reaching 30 million by 2100 or below 20 million in lower projections. Seoul’s population may reduce to one-third of its 2020 size by 2100. Persistently low fertility rates result in a high dependency ratio and accelerated aging, particularly in Seoul and Gyeonggi-do. Although old-age dependency ratios might improve slightly by 2100, economic challenges such as reduced purchasing power and socio-economic strain from an aging population and declining fertility remain significant. A probabilistic approach can enhance resource allocation strategies to support the aging population at both national and subnational levels. Full article

This paper focuses on electricity demand forecasting and its uncertainty representation using a hybrid machine learning (ML) model in the eastern control area of southeastern Mexico. In this case, different sources of uncertainty are integrated by applying the Bootstrap method, which adds the characteristics of stochastic noise, resulting in a hybrid probabilistic and ML model in the form of a time series. The proposed methodology addresses a function density probability, which is the generalized of extreme values obtained from the errors of the ML model; however, it is adaptable and independent and simulates the variability that may arise due to unforeseen events. Results indicate that for a five-day forecast using only demand data, the proposed model achieves a Mean Absolute Percentage Error (MAPE) of 4.358%; however, incorporating temperature increases the MAPE to 5.123% due to growing uncertainty. In contrast, a day-ahead forecast, including temperature, improves accuracy, reducing MAPE to 1.644%. The stochastic noise component enhances probabilistic modeling, yielding a MAPE of 3.042% with and 2.073% without temperature in five-day forecasts. Therefore, the proposed model proves useful for regions with high demand variability, such as southeastern Mexico, while maintaining accuracy over longer time horizons. Full article

This study explores the application of spiking neural networks (SNNs) for click-through rate (CTR) prediction in personalized online advertising systems, introducing a novel hybrid model, the Temporal Rate Spike with Attention Neural Network (TRA–SNN). By leveraging the biological plausibility and energy efficiency of SNNs, combined with attention-based mechanisms, the TRA–SNN model captures temporal dynamics and rate-based patterns to achieve performance comparable to state-of-the-art Artificial Neural Network (ANN)-based models, such as Deep & Cross Network v2 (DCN-V2) and FinalMLP. The models were trained and evaluated on the Avazu and Digix datasets, using standard metrics like AUC-ROC and accuracy. Through rigorous hyperparameter tuning and standardized preprocessing, this study ensures fair comparisons across models, highlighting SNNs’ potential for scalable, sustainable deployment in resource-constrained environments like mobile devices and large-scale ad platforms. This work is the first to apply SNNs to CTR prediction, setting a new benchmark for energy-efficient predictive modeling and opening avenues for future research in hybrid SNN–ANN architectures across domains like finance, healthcare, and autonomous systems. Full article

Youth unemployment remains a pressing issue in many emerging economies, where educational disparities and demographic pressures interact in complex ways. This study investigates the links between higher-education enrolment, demographic structure and youth unemployment in eight developing countries from 2009 to 2023. Panel cointegration techniques—Fully Modified Ordinary Least Squares (FMOLS) and Dynamic Ordinary Least Squares (DOLS)—are applied to estimate the long-run effects of gross tertiary-school enrolment on youth unemployment while controlling for GDP growth and youth-cohort size. Robustness is confirmed through complementary estimations with pooled-mean-group ARDL and system-GMM panels, which deliver consistent coefficient signs and significance levels. Results show a significant negative elasticity between enrolment and youth unemployment, indicating that wider access to higher education helps lower joblessness among young people. Youth-population growth exerts an opposite, positive effect, while GDP growth reduces unemployment but less uniformly across regions. The evidence points to an integrated policy mix—expanding tertiary (especially vocational and technical) education, managing demographic pressure and maintaining macro-economic stability—to improve youth-employment outcomes in emerging economies. Full article

This systematic literature review explores the application of artificial intelligence (AI) and machine learning (ML) in financial market forecasting, with a focus on four asset classes: equities, cryptocurrencies, commodities, and foreign exchange markets. Guided by the PRISMA methodology, the study identifies the most widely used predictive models, particularly LSTM, GRU, XGBoost, and hybrid deep learning architectures, as well as key evaluation metrics, such as RMSE and MAPE. The findings confirm that AI-based approaches, especially neural networks, outperform traditional statistical methods in capturing non-linear and high-dimensional dynamics. However, the analysis also reveals several critical research gaps. Most notably, current models are rarely embedded into real or simulated trading strategies, limiting their practical applicability. Furthermore, the sensitivity of widely used metrics like MAPE to volatility remains underexplored, particularly in highly unstable environments such as crypto markets. Temporal robustness is also a concern, as many studies fail to validate their models across different market regimes. While data covering one to ten years is most common, few studies assess performance stability over time. By highlighting these limitations, this review not only synthesizes the current state of the art but also outlines essential directions for future research. Specifically, it calls for greater emphasis on model interpretability, strategy-level evaluation, and volatility-aware validation frameworks, thereby contributing to the advancement of AI’s real-world utility in financial forecasting. Full article

Hyperparameter optimization (HPO) is critical for enhancing the predictive performance of machine learning models in credit risk assessment for peer-to-peer (P2P) lending. This study evaluates four HPO methods, Grid Search, Random Search, Hyperopt, and Optuna, across four models, Logistic Regression, Random Forest, XGBoost, and LightGBM, using three real-world datasets (Lending Club, Australia, Taiwan). We assess predictive accuracy (AUC, Sensitivity, Specificity, G-Mean), computational efficiency, robustness, and interpretability. LightGBM achieves the highest AUC (e.g., $70.77\%$ on Lending Club, $93.25\%$ on Australia, $77.85\%$ on Taiwan), with XGBoost performing comparably. Bayesian methods (Hyperopt, Optuna) match or approach Grid Search’s accuracy while reducing runtime by up to $75.7$-fold (e.g., $3.19$ vs. $241.47$ min for LightGBM on Lending Club). A sensitivity analysis confirms robust hyperparameter configurations, with AUC variations typically below $0.4\%$ under $\pm 10\%$ perturbations. A feature importance analysis, using gain and SHAP metrics, identifies debt-to-income ratio and employment title as key default predictors, with stable rankings (Spearman correlation $> 0.95, p<0.01$) across tuning methods, enhancing model interpretability. Operational impact depends on data quality, scalable infrastructure, fairness audits for features like employment title, and stakeholder collaboration to ensure compliance with regulations like the EU AI Act and U.S. Equal Credit Opportunity Act. These findings advocate Bayesian HPO and ensemble models in P2P lending, offering scalable, transparent, and fair solutions for default prediction, with future research suggested to explore advanced resampling, cost-sensitive metrics, and feature interactions. Full article

In this paper, multi-option probabilistic paired comparison models are presented and applied for prediction. As these models operate on the basis of probabilities, they can estimate the likelihood of future outcomes and thus predict future events. The aim of the paper is to demonstrate that these models have strong predictive capabilities when the information embedded into the data is properly utilized. To this end, we incorporate the degree (e.g., large or small) of the differences between the compared objects. By refining the usual three-option model, we define a five-option model capable of leveraging information derived from the goal differences. To incorporate additional information, the model is further extended to account for potential advantages in the comparisons. As a further refinement, temporal weighting is also introduced. These models are applied to forecasting football match outcomes in the top five European leagues (Premier League, La Liga, Serie A, Bundesliga, and Ligue 1), and their predictive performance is evaluated using various metrics. Based on the most recent football seasons, this model consistently delivers better predictive metrics, on average, than those of the already strong benchmark model. The effect of a home-field advantage is statistically supported across all five leagues. The model fits are illustrated using confidence intervals, and, as an interesting insight, we also present the evolution of the team strengths for the top four English clubs during the 2023/24 season. Full article

The pressure fluctuations generated during high-speed train passage through tunnels can compromise both the train’s structural integrity and passenger comfort, highlighting the need for the accurate prediction of external pressure wave amplitudes. To address the high computational cost of multi-condition Computational Fluid Dynamics simulations, this study proposes a hybrid method combining numerical simulation and machine learning. A dataset was generated using simulations with five input features: tunnel length, train length, train speed, blockage ratio, and measurement point location. Four machine learning models—random forest, support vector regression, Extreme Gradient Boosting, and Multilayer Perceptron (MLP)—were evaluated, with the MLP model showing the highest baseline accuracy. To further improve performance, six metaheuristic algorithms were applied to optimize the MLP model, among which, the sparrow search algorithm (SSA) achieved the highest accuracy, with R² = 0.993, MAPE = 0.052, and RMSE = 0.112. A SHapley Additive exPlanations (SHAP) analysis indicated that the train speed and the blockage ratio were the most influential features. This study provides an effective and interpretable method for pressure wave prediction in tunnel environments and demonstrates the first integration of SSA optimization into aerodynamic pressure modeling. Full article

We examine how prior specification affects the Bayesian Dirichlet Auto-Regressive Moving Average (B-DARMA) model for compositional time series. Through three simulation scenarios—correct specification, overfitting, and underfitting—we compare five priors: informative, horseshoe, Laplace, mixture of normals, and hierarchical. Under correct model specification, all priors perform similarly, although the horseshoe and hierarchical priors produce slightly lower bias. When the model overfits, strong shrinkage—particularly from the horseshoe prior—proves advantageous. However, none of the priors can compensate for model misspecification if key VAR/VMA terms are omitted. We apply B-DARMA to daily S&P 500 sector trading data, using a large-lag model to demonstrate overparameterization risks. Shrinkage priors effectively mitigate spurious complexity, whereas weakly informative priors inflate errors in volatile sectors. These findings highlight the critical role of carefully selecting priors and managing model complexity in compositional time-series analysis, particularly in high-dimensional settings. Full article

Financial fraud detection is a critical application area within the broader domains of cybersecurity and intelligent financial analytics. With the growing volume and complexity of digital transactions, the traditional rule-based and shallow learning models often fall short in detecting sophisticated fraud patterns. This study addresses the challenge of accurately identifying fraudulent financial activities, especially in highly imbalanced datasets where fraud instances are rare and often masked by legitimate behavior. The existing models also lack interpretability, limiting their utility in regulated financial environments. Experiments were conducted on three benchmark datasets: IEEE-CIS Fraud Detection, European Credit Card Transactions, and PaySim Mobile Money Simulation, each representing diverse transaction behaviors and data distributions. The proposed methodology integrates a transformer-based encoder, multi-teacher knowledge distillation, and a symbolic belief–desire–intention (BDI) reasoning layer to combine deep feature extraction with interpretable decision making. The novelty of this work lies in the incorporation of cognitive symbolic reasoning into a high-performance learning architecture for fraud detection. The performance was assessed using key metrics, including the F1-score, AUC, precision, recall, inference time, and model size. Results show that the proposed transformer–BDI model outperformed traditional and state-of-the-art baselines across all datasets, achieving improved fraud detection accuracy and interpretability while remaining computationally efficient for real-time deployment. Full article

In recent years, the prominence of probabilistic forecasting has risen among numerous research fields (finance, meteorology, banking, etc.). Best practices on using such forecasts are, however, neither well explained nor well understood. The question of the benefits derived from these forecasts is of primary interest, especially for the industrial sector. A sound methodology already exists to evaluate the value of probabilistic forecasts of binary events. In this paper, we introduce a comprehensive methodology for assessing the value of probabilistic forecasts of continuous variables, which is valid for a specific class of problems where the cost functions are piecewise linear. The proposed methodology is based on a set of visual diagnostic tools. In particular, we propose a new diagram called EVC (“Effective economic Value of a forecast of Continuous variable”) which provides the effective value of a forecast. Using simple case studies, we show that the value of probabilistic forecasts of continuous variables is strongly dependent on a key variable that we call the risk ratio. It leads to a quantitative metric of a value called the OEV (“Overall Effective Value”). The preliminary results suggest that typical OEVs demonstrate the benefits of probabilistic forecasting over a deterministic approach. Full article