Framework For Forecasting Research Articles

Innovative framework for accurate and transparent forecasting of energy consumption: A fusion of feature selection and interpretable machine learning

The study presents a novel framework integrating feature selection (FS) and machine learning (ML) techniques to forecast inland national energy consumption (EC) in the United Kingdom across all energy sources. This innovative framework strategically combines three FS approaches with five interpretable ML models using Shapley Additive Explanations (SHAP), with the dual goal of enhancing accuracy and transparency in EC predictions. By meticulously selecting the most pertinent features from diverse features—including meteorological conditions, socioeconomic parameters, and historical consumption patterns of different primary fuels—the proposed framework enhances the robustness of the forecasting model. This is achieved through benchmarking three FS approaches: ensemble filter, wrapper, and a hybrid ensemble filter-wrapper. In addition, we introduce a novel ensemble filter FS, synthesizing outcomes from multiple base FS methods to make well-informed decisions about feature retention. Experimental results underscore the efficacy of integrating both wrapper and ensemble filter-wrapper FS approaches with interpretable ML models, ensuring the forecasting process remains comprehensible and interpretable while utilizing a manageable number of features (four to eight). In addition, experimental results indicate that different feature subsets are usually selected for each combined FS approach and ML model. This study not only demonstrates the framework's capability to provide accurate forecasts but also establishes it as a valuable tool for policymakers and energy analysts.

Do ECB's rate hikes have spillover effects on the Hungarian BUBOR and the EUR/HUF exchange rate? A five‐variable VAR model approach using the Diebold‐Yilmaz spillover table

AbstractWe intend to show the directional volatility spillovers between the European short term interbank lending rates (3‐month Euro Interbank Offered Rate [EURIBOR] and Euro Short‐Term Rate [ESTR]) and the Hungarian Budapest Interbank Offered Rate (BUBOR) and Euro‐Hungarian Forint exchange rate. To determine the extent to which the variables affect each other's volatilities we build a five‐variable vector autoregression (VAR) and determine the spillover table like in Diebold‐Yilmaz's 2012 work. This methodology is preferred to a simple impulse response function (IRF) because we manage to avoid the problem of non‐orthogonal innovations via the generalized forecast error variance decomposition framework. The issue of variable ordering, therefore, does not arise. We focus on three episodes of increased volatility in Hungarian and European short‐term interest rates: Q3–Q4 of 2019, Q1 of 2020 and Q3 of 2022. These episodes correspond to volatility spikes in EU markets that to some extent had a measurable spillover effect on Hungarian interbank rates. We find that on average, across the entire sample of 957 observations, about 6.3% of the volatility forecast error variance in all five European and Hungarian variables comes from spillovers. The total and directional spillovers over the sample are extremely low. We conclude that the European Central Bank's surprise policy decisions have a marginal impact on Hungarian interbank rates. We also find that BUBOR is primarily a net receiver of spillovers from the MAX short‐term government bond benchmark rather than the EURIBOR—this disproved our initial considerations.

Prediction model of deep vein thrombosis risk after lower extremity orthopedic surgery

PurposeThis investigation was conceived to engineer and appraise a pioneering clinical nomogram, crafted to bridge the extant chasm in literature regarding the postoperative risk stratification for deep vein thrombosis (DVT) in the aftermath of lower extremity orthopedic procedures. This novel tool offers a sophisticated and discerning algorithm for risk prediction, heretofore unmet by existing methodologies. MethodsIn this retrospective observational study, clinical records of hospitalized patients who underwent lower extremity orthopedic surgery were collected at the Wuxi TCM Hospital Affiliated to the Nanjing University of Chinese Medicine between Jan 2017 and Oct 2019. The univariate and multivariate analysis with the backward stepwise method was applied to select features for the predictive nomogram. The performance of the nomogram was evaluated with respect to its discriminant capability, calibration ability, and clinical utility. ResultA total of 5773 in-hospital patients were eligible for the study, with the incidence of deep vein thrombosis being approximately 1 % in this population. Among 31 variables included, 5 of them were identified to be the predictive features in the nomogram, including age, mean corpuscular hemoglobin concentration (MCHC), D-dimer, platelet distribution width (PDW), and thrombin time (TT). The area under the receiver operating characteristic (ROC) curve in the training and validation cohort was 85.9 % (95%CI: 79.96 %–90.04 %) and 85.7 % (95%CI: 78.96 %–90.69 %), respectively. Both the calibration curves and decision curve analysis demonstrated the overall satisfactory performance of the model. ConclusionOur groundbreaking nomogram is distinguished by its unparalleled accuracy in discriminative and calibrating functions, complemented by its tangible clinical applicability. This innovative instrument is set to empower clinicians with a robust framework for the accurate forecasting of postoperative DVT, thus facilitating the crafting of bespoke and prompt therapeutic strategies, aligning with the rigorous standards upheld by the most esteemed biomedical journals.

Innovative framework for effective service parts management in the automotive industry

Effective service parts management and demand forecasting are crucial for optimizing operations in the automotive industry. However, existing literature lacks a comprehensive framework tailored to the specific context of the Thai automotive sector. This study addresses this gap by proposing a strategic approach to service parts management and demand forecasting in the Thai automotive industry. Drawing on a diverse set of methodologies, including classical time series models and advanced machine learning techniques, various forecasting models were assessed to identify the most effective approach for predicting service parts demand. Categorization of service parts based on demand criteria was conducted, and decision rules were developed to guide stocking strategies, balancing the need to minimize service disruptions with cost optimization. This analysis reveals substantial cost savings potential through strategic stocking guided by the developed decision rules. Furthermore, evaluation of the performance of different forecasting models recommends the adoption of Support Vector Regressor (SVR) as the most accurate model for forecasting service parts demand in this context. This research contributes to the automotive service industry by providing a nuanced framework for service parts management and demand forecasting, leading to cost-effective operations and enhanced service quality. The findings offer valuable insights for practitioners and policymakers seeking to improve efficiency and sustainability in the Thai automotive sector.

Assimilation of Geophysics-Derived Spatial Data for Model Calibration in Geologic CO2 Sequestration

Summary Uncertainty in geological models usually leads to large uncertainty in the predictions of risk-related system properties and/or risk metrics (e.g., CO2 plumes and CO2/brine leakage rates) at a geologic CO2 storage site. Different types of data (e.g., point measurements from monitoring wells and spatial data from 4D seismic surveys) can be leveraged or assimilated to reduce the risk predictions. In this work, we develop a novel framework for spatial data assimilation and risk forecasting. Under the U.S. Department of Energy’s National Risk Assessment Partnership (NRAP), we have developed a framework using an ensemble-based data assimilation approach for spatial data assimilation and forecasting. In particular, we took CO2 saturation maps interpreted from 4D seismic surveys as inputs for spatial data assimilation. Three seismic surveys at Years 1, 3, and 5 were considered in this study. Accordingly, three saturation maps were generated for data assimilation. The impact from the level of data noise was also investigated in this work. Our results show increased similarity between the updated reservoir models and the “ground-truth” model with the increased number of seismic surveys. Predictive accuracy in CO2 saturation plume increases with the increased number of seismic surveys as well. We also observed that with the increase in the level of data noise from 1% to 10%, the difference between the updated models and the ground truth does not increase significantly. Similar observations were made for the prediction of CO2 plume distribution at the end of the CO2 injection period by increasing the data noise.

MDG: A Multi-Task Dynamic Graph Generation Framework for Multivariate Time Series Forecasting

MDG: A Multi-Task Dynamic Graph Generation Framework for Multivariate Time Series Forecasting

Glu-Ensemble: An ensemble deep learning framework for blood glucose forecasting in type 2 diabetes patients

Diabetes is a chronic metabolic disorder characterized by elevated blood glucose levels, posing significant health risks such as cardiovascular disease, and nerve, kidney, and eye damage. Effective management of blood glucose is essential for individuals with diabetes to mitigate these risks. This study introduces the Glu-Ensemble, a deep learning framework designed for precise blood glucose forecasting in patients with type 2 diabetes. Unlike other predictive models, Glu-Ensemble addresses challenges related to small sample sizes, data quality issues, reliance on strict statistical assumptions, and the complexity of models. It enhances prediction accuracy and model generalizability by utilizing larger datasets and reduces bias inherent in many predictive models. The framework's unified approach, as opposed to patient-specific models, eliminates the need for initial calibration time, facilitating immediate blood glucose predictions for new patients. The obtained results indicate that Glu-Ensemble surpasses traditional methods in accuracy, as measured by root mean square error, mean absolute error, and error grid analysis. The Glu-Ensemble framework emerges as a promising tool for blood glucose level prediction in type 2 diabetes patients, warranting further investigation in clinical settings for its practical application.

Prediction and optimization of civil engineering material properties based on artificial intelligence

In the modern research field of predicting and optimizing the performance of civil engineering materials, the incorporation of artificial intelligence (AI) has provided a new perspective. The research is based on a thorough examination of the central role of AI technology and the application of different mathematical models in the task of material performance prediction and optimization, summarizing the current state of AI technology application in the field of civil engineering material performance prediction. Against this backdrop, the research further proposes a brand-new AI-based civil engineering material performance prediction method. This study, aiming to realize the optimization of material performance prediction tasks, utilized cutting-edge machine learning techniques and mathematical models, proposing this new AI prediction framework. Detailed discussions on design, implementation, and evaluation were conducted in the forecast framework, which includes a large number of tables, mathematical functions, and references. Through rigorous study and implementation, the prediction framework exhibited admirable performance, demonstrating extensive applicability for civil engineering material performance prediction and optimization. In conclusion, this new AI-based civil engineering material performance prediction method has demonstrated encouraging results. This study provides a new horizon, illustrating the tremendous potential and value of AI for predicting the performance of civil engineering materials, thereby improving efficiency and effectiveness in the civil engineering field. Overall, this new AI prediction method provides a fresh pathway for future research, making an essential contribution to anticipating and optimizing in civil engineering field.

MPSTAN: Metapopulation-Based Spatio-Temporal Attention Network for Epidemic Forecasting.

Accurate epidemic forecasting plays a vital role for governments to develop effective prevention measures for suppressing epidemics. Most of the present spatio-temporal models cannot provide a general framework for stable and accurate forecasting of epidemics with diverse evolutionary trends. Incorporating epidemiological domain knowledge ranging from single-patch to multi-patch into neural networks is expected to improve forecasting accuracy. However, relying solely on single-patch knowledge neglects inter-patch interactions, while constructing multi-patch knowledge is challenging without population mobility data. To address the aforementioned problems, we propose a novel hybrid model called metapopulation-based spatio-temporal attention network (MPSTAN). This model aims to improve the accuracy of epidemic forecasting by incorporating multi-patch epidemiological knowledge into a spatio-temporal model and adaptively defining inter-patch interactions. Moreover, we incorporate inter-patch epidemiological knowledge into both model construction and the loss function to help the model learn epidemic transmission dynamics. Extensive experiments conducted on two representative datasets with different epidemiological evolution trends demonstrate that our proposed model outperforms the baselines and provides more accurate and stable short- and long-term forecasting. We confirm the effectiveness of domain knowledge in the learning model and investigate the impact of different ways of integrating domain knowledge on forecasting. We observe that using domain knowledge in both model construction and the loss function leads to more efficient forecasting, and selecting appropriate domain knowledge can improve accuracy further.

Development of a probabilistic agricultural drought forecasting (PADF) framework under climate change

Drought has significant impacts on human survival and social development, particularly on crop production. Agricultural drought is the most direct consequence of drought on crops. In this study, a Probabilistic Agricultural Drought Forecasting (PADF) framework was developed to employ the Ensemble Bayesian Least Square Support Vector Machine (EBLSSVM) method for bias correction in precipitation and temperature projections from multiple Regional Climate Models (RCMs). Vine Copula-Based Projection Model (VCPM) was then developed for accurate agricultural drought projections, providing deterministic results and valuable 90 % predictive intervals. The results indicate that the EBLSSVM method can generate better climate projections than the original outputs from RCMs and bias-corrected results from other bias-correction techniques. Based on the projection results from VCPM, the study found that drought will be a significant concern in Fujian province, especially in the southeast coastal region. Drought conditions are projected to be more severe in the 2050s than in the 2080s, under both RCP4.5 and RCP8.5. The average SSI values during months with a wet trend ranged from 0.1 to 0.3, whereas months with a drought trend predominantly exhibited average SSI values exceeding -0.5. Notably, SSI values as low as -2.0 were observed during wet trend months, underscoring the urgency of addressing future drought, particularly in coastal regions. However, even during wet periods, at least one extreme drought month is expected, suggesting that extreme drought conditions will become more severe in the future. CMIP5 and CMIP6 predictions showed good consistency in temporal and spatial dimensions, with CMIP6 indicating more significant and consistent future drought changes compared to CMIP5.

Platinum and palladium price forecasting through neural networks

To many commodity market participants, forecasts of price series represent a critical task. In this work, nonlinear autoregressive neural network models’ potential is explored for forecasting daily prices series of platinum and palladium over about a fifty-year period. For this purpose, one hundred and twenty model settings are examined, including different training algorithms, numbers of hidden neurons and delays, and ratios used to segment the data. With the analysis, two models leading to stable and accurate forecast results are constructed for the prices of the two commodities. In particular, the models’ performance in terms of the relative root mean square error is 1.86% and 3.61% for platinum and palladium, respectively, for the overall sample. Results in this work could help technical forecasts and policy analysis. The forecast framework might be extended to other different commodities.

Improved collaborative filtering for cross-store demand forecasting

Accurate and reliable demand forecasting is crucial for decision-making in assortment planning, store location, and various operational aspects. This study proposes a hybrid TS-SimPMF framework for multi-store multi-product demand forecasting, employing probabilistic matrix factorization (PMF) with time-series similarity measures to capture product and store heterogeneity. The heterogeneity enables the framework to achieve multi-step demand forecasting under unknown item-store combinations through an adaptive integration strategy. We first employ a hybrid dynamic regularization approach to measure the similarity of product sales simultaneously, and then the PMF framework is applied to generate latent vectors of products and stores. The performance of the proposed framework is verified by the real-world retail data, in which the dataset includes all sales records with 50 SKUs in 10 locations, totaling over 180,000 sales records. These experimental results demonstrate the superior performance of the proposed hybrid model compared to state-of-the-art methods.

GCN–Informer: A Novel Framework for Mid-Term Photovoltaic Power Forecasting

Predicting photovoltaic (PV) power generation is a crucial task in the field of clean energy. Achieving high-accuracy PV power prediction requires addressing two challenges in current deep learning methods: (1) In photovoltaic power generation prediction, traditional deep learning methods often generate predictions for long sequences one by one, significantly impacting the efficiency of model predictions. As the scale of photovoltaic power stations expands and the demand for predictions increases, this sequential prediction approach may lead to slow prediction speeds, making it difficult to meet real-time prediction requirements. (2) Feature extraction is a crucial step in photovoltaic power generation prediction. However, traditional feature extraction methods often focus solely on surface features, and fail to capture the inherent relationships between various influencing factors in photovoltaic power generation data, such as light intensity, temperature, and more. To overcome these limitations, this paper proposes a mid-term PV power prediction model that combines Graph Convolutional Network (GCN) and Informer models. This fusion model leverages the multi-output capability of the Informer model to ensure the timely generation of predictions for long sequences. Additionally, it harnesses the feature extraction ability of the GCN model from nodes, utilizing graph convolutional modules to extract feature information from the ‘query’ and ‘key’ components within the attention mechanism. This approach provides more reliable feature information for mid-term PV power prediction, thereby ensuring the accuracy of long sequence predictions. Results demonstrate that the GCN–Informer model significantly reduces prediction errors while improving the precision of power generation forecasting compared to the original Informer model. Overall, this research enhances the prediction accuracy of PV power generation and contributes to advancing the field of clean energy.

PaSTG: A Parallel Spatio-Temporal GCN Framework for Traffic Forecasting in Smart City

Predicting future traffic conditions from urban sensor data is crucial for smart city applications. Recent traffic forecasting methods are derived from Spatio-Temporal Graph Convolution Networks (STGCNs). Despite their remarkable achievements, these spatio-temporal models have mainly been evaluated on small-scale datasets. In light of the rapid growth of the Internet of Things and urbanization, cities are witnessing an increased deployment of sensors, resulting in the collection of extensive sensor data to provide more accurate insights into citywide traffic dynamics. Spatio-temporal graph modeling on large-scale traffic data is challenging due to the memory constraint of the computing device. For traffic forecasting, subgraph sampling from road networks onto multiple devices is feasible. Many GCN sampling methods have been proposed recently. However, combining these with STGCNs degrades performance. This is primarily due to prediction biases introduced by each sampled subgraph, which analyze traffic states from a regional perspective. Addressing these challenges, we introduce a parallel STGCN framework called PaSTG. PaSTG divides the road network into regions, each processed by an individual STGCN in a device. To mitigate regional biases, Aggregation Blocks in PaSTG merge spatial-temporal features from each STBlock. This collaboration enhances traffic forecasting. Furthermore, PaSTG implements pipeline parallelism and employs a graph partition algorithm for optimized pipeline efficiency. We evaluate PaSTG on various STGCNs using three traffic datasets on multiple GPUs. Results demonstrate that our parallel approach applies widely to diverse STGCN models, surpassing existing GCN samplers by up to 57.4% in prediction accuracy. Additionally, the parallel framework achieves speedups of up to 2.87x and 4.70x in training and inference compared to GCN samplers.

Learning evolving relations for multivariate time series forecasting

Multivariate time series forecasting is essential in various fields, including healthcare and traffic management, but it is a challenging task due to the strong dynamics in both intra-channel relations (temporal patterns within individual variables) and inter-channel relations (the relationships between variables), which can evolve over time with abrupt changes. This paper proposes ERAN (Evolving Relational Attention Network), a framework for multivariate time series forecasting, that is capable to capture such dynamics of these relations. On the one hand, ERAN represents inter-channel relations with a graph which evolves over time, modeled using a recurrent neural network. On the other hand, ERAN represents the intra-channel relations using a temporal attentional convolution, which captures the local temporal dependencies adaptively with the input data. The elvoving graph structure and the temporal attentional convolution are intergrated in a unified model to capture both types of relations. The model is experimented on a large number of real-life datasets including traffic flows, energy consumption, and COVID-19 transmission data. The experimental results show a significant improvement over the state-of-the-art methods in multivariate time series forecasting particularly for non-stationary data.

A general explicable forecasting framework for weather events based on ordinal classification and inductive rules combined with fuzzy logic

This paper presents a method for providing explainability in the integration of artificial intelligence (AI) and data mining techniques when dealing with meteorological prediction. Explainable artificial intelligence (XAI) refers to the transparency of AI systems in providing explanations for their predictions and decision-making processes, and contribute to improve prediction accuracy and enhance trust in AI systems. The focus of this paper relies on the interpretability challenges in ordinal classification problems within weather forecasting. Ordinal classification involves predicting weather phenomena with ordered classes, such as temperature ranges, wind speed, precipitation levels, and others. To address this challenge, a novel and general explicable forecasting framework, that combines inductive rules and fuzzy logic, is proposed in this work. Inductive rules, derived from historical weather data, provide a logical and interpretable basis for forecasting; while fuzzy logic handles the uncertainty and imprecision in the weather data. The system predicts a set of probabilities that the incoming sample belongs to each considered class. Moreover, it allows the expert decision-making process to be strengthened by relying on the transparency and physical explainability of the model, and not only on the output of a black-box algorithm. The proposed framework is evaluated using two real-world weather databases related to wind speed and low-visibility events due to fog. The results are compared to both ML classifiers and specific methods for ordinal classification problems, achieving very competitive results in terms of ordinal performance metrics while offering a higher level of explainability and transparency compared to existing approaches.

Exploring the potentials of artificial intelligence towards carbon neutrality: Technological convergence forecasting through link prediction and community detection

To explore future opportunities for artificial intelligence (AI) in achieving carbon neutrality goals, this study introduces a forecasting framework that combines link prediction and community detection. The proposed framework begins by utilizing patent information and the International Patent Classification to construct a co-occurrence network. Subsequently, it is computed and validated using two emerging technology datasets: one related to AI towards carbon neutrality and the other to Virtual Reality. A feature selection method and classification model are employed in this process. Finally, predicted convergence links are clustered using network modularity detection to unveil future trends. Through a comprehensive comparison of various similarity indicators in link prediction, this study identifies local indices (Resource Allocation, Adam-Adar) and global indices (Rooted Pagerank, SimRank) as the most featured prediction indicators. These research findings address a gap in the existing literature by providing insights into the rationale behind the selection of link similarity indicators, thereby resolving the issue of randomness in indicator selection. This study also finds that among all single and ensemble models, tree-based models, particularly Random Forest, emerge as the best-performing classifiers. The predicted results suggest that AI technologies towards carbon neutrality can be categorized into two future trends: theoretical methodological foundations and application sectors. This enables the identification and recommendation of future technological potentials on both micro and macro levels. Specifically, it facilitates the discovery of the most promising subdivision technology fields in the future, operating at the microlevel of patent main grouping, often characterized by significant cross-disciplinary links, and at the macrolevel, by visualizing potential technological convergence networks and clustering the trends. The analytical processes and quantitative outcomes of this study can support strategic decision-making for AI, offering valuable insights for both policy-makers and practitioners in their pursuit of green innovation solutions towards carbon neutrality.

Interpretable corn future price forecasting with multivariate time series

AbstractEfforts in corn future price forecasting and early warning play a vital role in guiding the high‐quality development of the agricultural economy. However, recent years have witnessed significant fluctuations in global corn future prices due to the impact of COVID‐19 and the escalating risks associated with geopolitical conflicts. Therefore, there is an urgent need for accurate and efficient methods to forecast corn future prices. To address this challenge, a novel and comprehensive framework for explainable corn future price forecasting is designed. This framework takes into account multiple factors contributing to corn price volatility, including supply and demand dynamics, policy adjustments, international market shocks, global geopolitical risks, and investor concerns within the corn market. During the data processing stage, complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is utilized to thoroughly explore the volatility characteristics of historical corn future prices. Additionally, a convolutional neural network (CNN) is employed to extract essential forecasting information from corn news data. To enhance interpretability, a novel JADE–TFT interpretable corn future price prediction model is proposed. This model combines adaptive differential evolution with optional external archiving (JADE) to intelligently and efficiently optimize the parameters of the temporal fusion transformers (TFTs). Furthermore, in the empirical study, the introduction of a global geopolitical risk coefficient, Baidu indices such as “corn” and “corn price,” and quantized corn news text features is shown to improve the accuracy of corn future price predictions. The proposed corn future price prediction framework contributes to the healthy development of the global grain futures market, thereby fostering the growth and well‐being of enterprises involved in the grain industry.

BO–SHAP–BLS: a novel machine learning framework for accurate forecasting of COVID-19 testing capabilities

BO–SHAP–BLS: a novel machine learning framework for accurate forecasting of COVID-19 testing capabilities

A bi-level framework for real-time crash risk forecasting using artificial intelligence-based video analytics.

This study proposes a bi-level framework for real-time crash risk forecasting (RTCF) for signalised intersections, leveraging the temporal dependency among crash risks of contiguous time slices. At the first level of RTCF, a non-stationary generalised extreme value (GEV) model is developed to estimate the rear-end crash risk in real time (i.e., at a signal cycle level). Artificial intelligence techniques, like YOLO and DeepSort were used to extract traffic conflicts and time-varying covariates from traffic movement videos at three signalised intersections in Queensland, Australia. The estimated crash frequency from the non-stationary GEV model is compared against the historical crashes for the study locations (serving as ground truth), and the results indicate a close match between the estimated and observed crashes. Notably, the estimated mean crashes lie within the confidence intervals of observed crashes, further demonstrating the accuracy of the extreme value model. At the second level of RTCF, the estimated signal cycle crash risk is fed to a recurrent neural network to predict the crash risk of the subsequent signal cycles. Results reveal that the model can reasonably estimate crash risk for the next 20-25min. The RTCF framework provides new pathways for proactive safety management at signalised intersections.