Forecast Accuracy Research Articles

Assessing forecast performance of daily reference evapotranspiration: A comparison of equations, machine and deep learning using weather forecasts

An accurate forecast of short-term reference evapotranspiration (ET0) is crucial for effective farm and irrigation scheduling management. The techniques of forecasting ET0 have progressed from empirical equations to artificial intelligence-based models. However, gaps have remained in comprehensively evaluating the ET0 forecast performance discrepancies between various methods based on weather forecasts in addition to historical meteorological data, as well as investigating the effect of the amount of input variables on ET0 forecast accuracy. Here, the study evaluated the forecast accuracy of five conventional equations calibrated and six machine learning (ML) and deep learning (DL) models in predicting daily ET0 for a lead time of 1–7 days in the North China Plain (NCP). Comparative evaluations indicate that the Analytical Penman-Monteith (APM) is the optimum empirical equation for forecasting ET0, but its accuracy decreases as lead times increase. Conversely, ML and DL models consistently outperform empirical equations, characterized by lower mean absolute error (MAE) and root mean square error (RMSE) as well as correlation coefficient and Nash-Sutcliffe efficiency coefficient approaching 1 across all lead times. Notably, the Bidirectional Long short-term memory (Bi-LSTM) model outperforms other models, maintaining robust and effective forecast performance even on day 7. Moreover, with the increasing input variables, combining weather forecasts and historical meteorological data, the forecast performance of all models is improved, with significant enhancement observed in the Bi-LSTM model. This work provides valuable insights for choosing appropriate models to forecast ET0, holding essential in regional managing farm and irrigation systems.

GWO-ANFIS-RD3PG: A reinforcement learning approach with dynamic adjustment and dual replay mechanism for building energy forecasting

Reliable prediction of building energy consumption is essential for effective and sustainable energy management. Traditional forecasting methods, however, face challenges when dealing with sudden changes in energy consumption. To address the above-mentioned issue, we introduce the GWO-ANFIS-RD3PG prediction model in this paper. This model aims to ensure global forecasting accuracy while minimizing prediction errors at sudden change points. A Grey Wolf Optimization (GWO) algorithm with the Adaptive Neuro-Fuzzy Inference System (ANFIS) is proposed to predict the energy consumption type and its fluctuation magnitude for the next time step. Within the Recurrent Dual Experience Replay Buffer DDPG (RD3PG) prediction module, we employ Long Short-Term Memory (LSTM), as the actor network in the Deep Deterministic Policy Gradient (DDPG), to account for long-term dependencies in time series data. Notably, we introduce an adaptive action modification mechanism that allows the agent to optimize actions based on the output of the aforementioned ANFIS, enabling real-time adjustments during sudden change. To further enhance the model’s performance, we adopt a dual experience replay mechanism to store data samples from sudden change points, capturing insight information during these anomalies. Experimental results on two real-world datasets demonstrate that this method reduces MAE by 5.81% and 13.47%, and RMSE by 5.69% and 15.21% compared to the state-of-art models, showing the significance of the proposed method for energy efficiency improvement in real-world building operations.

Multi-Step-Ahead Time Series Forecasting using Deep Learning and Fuzzy Time Series-based Error Correction Method

Recently time series forecasting has become one of the prime application areas of climatology, economics and industries. Many research works are conducted to forecast the time series more accurately. But few of them are concentrated on predicting the time series over an extended future horizon, and there is also a scope to improve their forecasting accuracy. This work proposes a multi-step-ahead foresting method to produce a stable and accurate forecasting result for the extended future horizon. Firstly, a deep learning-based forecasting model is proposed to predict the time series. Secondly, a fuzzy time series-based error correction model is implemented to enhance the prediction performance of the deep learning model. Here to optimize all the fuzzy time series (FTS) parameters in an integrated way, an integrated butterfly optimization (FTS-IBO) algorithm is proposed. In this study, two different types of real-world multivariate time series datasets are used to analyze the forecasting performance of the proposed model. The performance of the proposed FTS-IBO algorithm is compared with the traditional butterfly optimization (FTS-BO) algorithm. The experimental results show that the FTS-IBO technique is superior to the FTS-BO technique. The forecasting performance of the proposed model has also compared the other state-of-the-art models, and the simulation results exhibit that the proposed model produces a more accurate prediction performance for multi-step-ahead time series forecasting problems compared to other models.

Optimizing Multivariate Time Series Forecasting with Data Augmentation

The convergence of data mining and deep learning has become an invaluable tool for gaining insights into evolving events and trends. However, a persistent challenge in utilizing these techniques for forecasting lies in the limited access to comprehensive, error-free data. This challenge is particularly pronounced in financial time series datasets, which are known for their volatility. To address this issue, a novel approach to data augmentation has been introduced, specifically tailored for financial time series forecasting. This approach leverages the power of Generative Adversarial Networks to generate synthetic data that replicate the distribution of authentic data. By integrating synthetic data with real data, the proposed approach significantly improves forecasting accuracy. Tests with real datasets have proven that this method offers a marked improvement over models that rely only on real data.

Identifying Financially Material Content in Corporate Social Responsibility Reports

ABSTRACTThis study examines the relationship between financially material content in corporate social responsibility (CSR) reports and the decision usefulness of these reports. Utilizing sustainability disclosure standards and a machine learning topic modeling algorithm, a firm‐specific quantitative measure of financially material content in CSR reports is developed. It is hypothesized that firms providing greater amounts of financially material CSR content enhance their information environment, which enables analysts to make more accurate earnings predictions. The findings confirm a positive relationship between the extent of financially material CSR content disclosed and analyst forecast accuracy. This research demonstrates the effectiveness of using machine learning to identify financially material content within unstructured voluntary disclosures and contributes to the literature on the financial materiality of CSR activities and their related disclosures.

Long-Term Passenger Flow Forecasting for Rail Transit Based on Complex Networks and Informer

With the continuous growth of urbanization, passenger flow in urban rail transit systems is steadily increasing, making accurate long-term forecasting essential for optimizing operational scheduling and enhancing service quality. However, passenger flow forecasting becomes increasingly complex due to the intricate structure of rail transit networks and external factors such as seasonal variations. To address these challenges, this paper introduces an optimized Informer model for long-term forecasting that incorporates the influences of other stations based on complex network theory. Compared to the ARIMA, LSTM, and Transformer models, this optimized Informer model excels in processing large-scale complex transit data, particularly in terms of long-term forecasting accuracy and capturing network dependencies. The results demonstrate that this forecasting approach, which integrates complex network theory with the Informer model, significantly improves the accuracy and efficiency of long-term passenger flow predictions, providing robust decision support for urban rail transit planning and management.

Enhanced Short-Term Load Forecasting: Error-Weighted and Hybrid Model Approach

To tackle the challenges of high variability and low accuracy in short-term electricity load forecasting, this study introduces an enhanced prediction model that addresses overfitting issues by integrating an error-optimal weighting approach with an improved ensemble forecasting framework. The model employs a hybrid algorithm combining grey relational analysis and radial kernel principal component analysis to preprocess the multi-dimensional input data. It then leverages an ensemble of an optimized deep bidirectional gated recurrent unit (BiGRU), an enhanced long short-term memory (LSTM) network, and an advanced temporal convolutional neural network (TCN) to generate predictions. These predictions are refined using an error-optimal weighting scheme to yield the final forecasts. Furthermore, a Bayesian-optimized Bagging and Extreme Gradient Boosting (XGBoost) ensemble model is applied to minimize prediction errors. Comparative analysis with existing forecasting models demonstrates superior performance, with an average absolute percentage error (MAPE) of 1.05% and a coefficient of determination (R2) of 0.9878. These results not only validate the efficacy of our proposed strategy, but also highlight its potential to enhance the precision of short-term load forecasting, thereby contributing to the stability of power systems and supporting societal production needs.

Numerical experiment on data assimilation for geothermal doublets using production data and electromagnetic observations

The data assimilation process for geothermal reservoirs often relies on well data, which primarily offer insights into the immediate vicinity of the borehole. However, integrating geophysical methods can provide valuable information beyond well proximity, possibly enhancing reservoir predictions. Current methods of monitoring geothermal reservoirs struggle to maintain a good signal-to-noise ratio for deep reservoirs. Diffusive electromagnetic (EM) methods can be sensitive to the decreasing conductivity from heat extraction in geothermal reservoirs and offer promising additional value. To test their potential effectiveness, numerical examples are simulated. A scheme to incorporate diffusive EM observations into a data assimilation process for geothermal reservoirs is presented and implemented in this study. First, an ensemble of prior models representing the reservoir uncertainty is used to determine the moments of the resulting temperature field using a forward geothermal simulation. Subsequently, a conductivity model is calculated from the temperature field using an empirical relationship. The expected electric field response can then be simulated using an EM forward model. EM sources are placed on the surface around the expected cold plume location. The receiver is placed at reservoir depth. To assimilate the data, the ensemble smoother with multiple data assimilation method is used. The findings demonstrate that the incorporation of EM data provides valuable information regarding the temperature field. This improves the accuracy of the temperature forecast of the entire reservoir when combined with the localized data from the production well and, therefore, helps to resolve the complex migration of the cold front. These results highlight the monitoring potential of EM observations for geothermal reservoirs.

Development of PM2.5 Forecast Model Combining ConvLSTM and DNN in Seoul

Accurate prediction of PM2.5 concentrations is essential for public health management, especially in areas affected by long-range pollutant transport. This study presents a hybrid model combining convolutional long short-term memory (ConvLSTM) and deep neural networks (DNNs) to enhance PM2.5 forecasting in Seoul, South Korea. The hybrid model leverages ConvLSTM’s ability to capture spatiotemporal dependencies and DNN’s strength in feature extraction, enabling it to outperform standalone CMAQ and DNN models. For the T1 forecast (6 h averages), the ConvLSTM-DNN model exhibited superior performance, with an RMSE of 7.2 µg/m3 compared to DNN’s 8.5 µg/m3 and CMAQ’s 10.1 µg/m3. The model also maintained high categorical accuracy (ACC) and probability of detection (POD) for critical PM2.5 levels while reducing false alarms (FARs), particularly in bad and very bad events. Although its performance decreases over extended forecast periods, the ConvLSTM-DNN model demonstrates its utility as a robust forecasting tool. Future work will focus on optimizing the network structure to improve long-term forecast accuracy.

High-resolution spatiotemporal forecasting of the European crane migration

In this paper we present three different models to forecast bird migration. They are species-specific individual-based models that operate on a high spatiotemporal resolution (kilometres, 15 min-hours), as an addition to radar-based migration forecast models that currently exist. The models vary in complexity, and use GPS-tracked location, flying direction and speed, and/or wind data to forecast migration speed and direction. Our aim is to quantitatively evaluate the forecasting performance and assess which metrics improve forecasts at different ranges. We test the models through cross-validation using GPS tracks of common cranes during spring and autumn migration. Our results show that recordings of flight speed and direction improve the accuracy of forecasts on the short range (<2 h). Adding wind data at flight altitude results in consistent improvements of the forecasts across the entire range, particularly in the predicted speed. Direction forecasts are less affected by adding wind data because cranes mostly compensate for wind drift during migration. Migration in spring is more difficult to forecast than in autumn, resulting in larger errors in flight speed and direction during spring. We further find that a combination of flight behaviours – thermal soaring, gliding, and flapping – complicates the forecasts by inducing variance in flight speed and direction. Fitting those behaviours into flight optimisation models proves to be challenging, and even results in significant biases in speed forecasts in spring. We conclude that flight speed is the most difficult parameter to forecast, whereas flight direction is the most critical for practical applications of these models. Such applications could e.g., be prevention of bird strikes in aviation or with wind turbines, and public engagement with bird migration.

Integrated Systematic Framework for Forecasting China’s Consumer Confidence: A Machine Learning Approach

This study aims to introduce a novel approach for predicting China’s consumer confidence index (CCI), a key economic indicator that reflects consumers’ confidence in current and future economic conditions. While traditional statistical models and economic indicators are the primary tools for forecasting CCI, their reliance on linear assumptions limits their ability to capture the complex, dynamic relationships inherent in economic systems. In response, this study proposes a two-step method that integrates social network analysis (SNA) and machine learning (ML) to enhance prediction accuracy by accounting for the nonlinear interactions and systemic interdependencies that drive consumer confidence. The use of SNA enables the identification of critical variables and their interconnected roles in shaping consumer sentiment, while ML models, specifically the gradient boosting decision tree (GBDT), leverage these relationships to provide more precise predictions. Utilizing monthly data from 1999 to 2023, the combined SNA and GBDT approach significantly improves the accuracy of CCI forecasts, particularly during periods of high volatility. The results of this study hold substantial value for policymakers, market analysts, and economists, as they offer a systems-oriented framework for economic forecasting. By demonstrating the effectiveness of combining SNA with ML technologies, this research not only advances the methodological toolkit for economic forecasting, but also provides a new lens through which the complex, adaptive nature of economic systems can be better understood and managed. This integrated approach paves the way for future developments in forecasting models that more accurately reflect the evolving dynamics of consumer confidence in a rapidly changing economic environment.

Testing for Equal Average Forecast Accuracy in Possibly Unstable Environments *

We consider the issue of testing the null of equal average forecast accuracy in a model where the forecast error loss differential series has a potentially non-constant mean function over time. We show that when time variation is present in the loss differential mean, the standard Diebold and Mariano (1995) test, which was proposed for evaluating forecasts in a stable environment, has an asymptotic size of zero, and, whilst consistent, can have reduced local power. This arises due to inconsistent estimation of the implicit long run variance estimator, which diverges under a time varying mean. We suggest a modified statistic that replaces the standard long run variance estimator based on full-sample demeaning of the loss differential series with one based on nonparametric local demeaning. The new long run variance estimator is consistent under both the null and alternative when the mean function is time varying or constant, and in both cases, the modified test recovers the asymptotic size and power properties associated with the original test in the constant mean case. The modified test therefore provides a robust method for testing the equal average forecast accuracy null, allowing for instability in the loss differential mean. The benefits of our test are demonstrated via Monte Carlo simulation and two empirical applications.

Forecasting Orange Juice Futures: LSTM, ConvLSTM, and Traditional Models Across Trading Horizons

This study evaluated the forecasting accuracy of various models over 5-day and 10-day trading horizons to predict the prices of orange juice futures (OJ = F). The analysis included traditional models like Autoregressive Integrated Moving Average (ARIMA) and advanced neural network models such as Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), Backpropagation Neural Network (BPNN), Support Vector Regression (SVR), and Convolutional Long Short-Term Memory (ConvLSTM), incorporating factors like the Commodities Index and the S&amp;P500 Index. We employed loss function metrics and various tests to assess model performance. The results indicated that for the 5-day horizon, the LSTM and ConvLSTM consistently outperformed the other models. LSTM achieved the lowest error rates and demonstrated superior capability in capturing temporal dependencies, especially in single-factor and S&amp;P500 Index predictions. ConvLSTM also performed strongly, effectively modeling spatial and temporal data patterns. In the 10-day horizon, similar trends were observed. LSTM and ConvLSTM models had significantly lower errors and better alignment with actual values. The BPNN model performed well when all factors were included, and the SVR model maintained consistent accuracy, particularly for single-factor predictions. The Diebold–Mariano (DM) test indicated significant differences in forecasting accuracy, favoring advanced neural network models. In addition, incorporating multiple influencing factors further improved predictive performance, enhancing investment outcomes and reducing risk.

Tsunami Early Warning Using High-Frequency Ocean Radar System in the Kii Channel, Japan

Abstract The tsunami of the 2011 Tohoku earthquake was monitored by the high-frequency (HF) ocean radar system in the Kii Channel, Japan. In this study, we conducted a retrospective analysis of tsunami early warning based on HF radar observations. First, we synthesized the surface current velocity using observation data of two radar land stations: Minato and Saikazaki. After vector synthesis, the time series of current velocity showed high consistency with the results of forward simulation. Then, we used the time series starting from 08:00 (UTC), 11 March 2011, as the input for data assimilation, reconstructing the tsunami wavefield in the Kii Channel. Comparison with actual observations at Kobe showed that our approach can accurately forecast coastal tsunami waveforms at least 50 min prior to tsunami arrival. The normalized prediction error decreased over time, indicating a progressively improving forecast accuracy. This is the first study on tsunami data assimilation using HF radar for the early warning of an actual seismogenic tsunami. We also discussed the limitations of the current HF radar system in the Kii Channel and made suggestions regarding the deployment of future radar land stations.

Improving out-of-sample forecasts of stock price indexes with forecast reconciliation and clustering

In this paper, we propose a novel approach to improving forecasts of stock market indexes by considering common stock prices as hierarchical time series, combining clustering with forecast reconciliation. We propose grouping the individual stock price series in various ways including via metadata and using unsupervised learning techniques. The proposed approach is applied to the Dow Jones Industrial Average Index and the Standard & Poor 500 Index and their component stocks, and the results obtained with different grouping approaches are compared. The results empirically demonstrate that the combined use of clustering and reconciliation improves the forecast accuracy of the stock market indexes and their constituents.

Introducing edge intelligence to smart meters via federated split learning

The ubiquitous smart meters are expected to be a central feature of future smart grids because they enable the collection of massive amounts of fine-grained consumption data to support demand-side flexibility. However, current smart meters are not smart enough. They can only perform basic data collection and communication functions and cannot carry out on-device intelligent data analytics due to hardware constraints in terms of memory, computation, and communication capacity. Moreover, privacy concerns have hindered the utilization of data from distributed smart meters. Here, we present an end-edge-cloud federated split learning framework to enable collaborative model training on resource-constrained smart meters with the assistance of edge and cloud servers in a resource-efficient and privacy-enhancing manner. The proposed method is validated on a hardware platform to conduct building and household load forecasting on smart meters that only have 192 KB of static random-access memory (SRAM). We show that the proposed method can reduce the memory footprint by 95.5%, the training time by 94.8%, and the communication burden by 50% under the distributed learning framework and can achieve comparable or superior forecasting accuracy to that of conventional methods trained on high-capacity servers.

An Integrated Approach for Development of a Decision Support System for Early Surveillance of Japanese Encephalitis: a Remote Sensing-based study in Conjunction with Epidemiological Scenario in Assam, India

Abstract. Japanese Encephalitis (JE) is one of the serious mosquito borne viral diseases mostly prevalent in India and many of the South Asian countries. A comprehensive study was conducted in Dibrugarh, a JE endemic district of Assam taking into the account of the environmental and social factors that determine the outbreak of the disease. Effect of environmental and climatic factors on mosquito vector densities was investigated using geospatial tools and techniques. Wetlands were found to be the most preferred habitat type for most of the JE-transmitting mosquitoes followed by paddy-growing areas and ponds. The villages abundant with these habitats or located nearby were found to be more vulnerable to JE risk. Based on the spatial distribution of habitat types, villages in the district were divided into three categories of vector abundance and hence the risk of JE. It is found that rainfall (intensity as well as duration) is the most significant determinant in the modulation of mosquito density followed by temperature. It is also observed that the derived models could explain about 61% to 73% variation in mosquito density due to change in weather variables as indicated by R2 values of the Models. Among socio-economic determinants, pig rearing habit of the villagers was found to be most significant. JE risk map was prepared by integrating vector abundance map with the map of host abundance in GIS domain. Attempts were also made to forecast the intensity of JE cases based on a time series analysis of historical morbidity pattern, whereas forecast for onset of the disease was based on the modulation of weather variables in the study area. A web-based decision support system (DSS) developed by integrating the forecasts of JE onset, intensity and JE prone villages has been provided to the concerned health authorities to take timely intervention measures. The JE early warning system developed for Dibrugarh district was extended to two more JE prone districts viz, Sibsagar and Tinsukia as validation sites and observed that the accuracy of forecast was within the range of 69% to 77%. DSS on JE has opened up new opportunities to study similar vector-borne diseases of the JE serogroup viz., West Nile.

Основные особенности климатических условий зимнего сезона 2023/2024 гг. по данным мониторинга и прогнозов

The main results of the comprehensive analysis of the Northern Hemisphere large-scale atmospheric circulation features are presented for the 2023/2024 winter season. Skill scores of the consensus forecast for the 2023/2024 Northern Eurasia winter season are discussed in the context of reproducing the temperature and humidity regime. The qualitative analysis of the multimodel forecast from WMO’s Lead Center and the consensus forecast issued by NEACOF for seasonal anomalies in air temperature and precipitation for the winter of 2023/2024 resulted in the conclusion of the superior accuracy of the consensus forecast. This forecast was based on the data of three Russian models SL-AV, MGO, and INM with equal weighting coefficients.

A hybrid load forecasting system based on data augmentation and ensemble learning under limited feature availability

Accurate power load forecasting is an important part of power system operation planning, it can ensure the stable operation of power systems and improve the efficiency of energy utilization. The power load is affected by many factors including temperature, season, population density, and so on, however due to privacy protection and other reasons, it is difficult to obtain some characteristic information that affects the load. The lack of characteristic data will reduce the accuracy of load forecasting and the generalization ability. To solve it, a new hybrid load forecasting framework is proposed, which is composed of two subsystems: a data preprocessing system and a high-precision forecasting system. Based on the load sequence itself, subsystem 1 obtains the trend data and denoising data by variational mode decomposition method, obtains the indicator variable for the weekend according to the one-hot encoding, and also introduces the electricity price data, thus obtaining the 4-dimensional extended data. Subsystem 2 constructs a hybrid prediction model by synthesizing various models, including deep learning and machine learning models, to forecast the expanded data. Finally, the multi-objective JAYA algorithm based on tent chaotic mapping and cross-perturbation strategy is used to ensemble the prediction results of the sub-models. To verify the superiority of the proposed forecasting framework, we conducted experiments using four sets of load data from New South Wales, Australia. The experimental results show that the average absolute percentage error of the hybrid framework on the four data sets are MAPEMarch=0.8070, MAPEJune=0.8296, MAPESeptember=0.7238 and MAPEDecember=0.7709, which are significantly lower than other models and provide a basis for power system scheduling management.

Transferability of models for predicting potato plant nitrogen content from remote sensing data and environmental variables across years and regions

The use of remote sensing technologies to monitor the nitrogen nutrient status of crops is gradually becoming a more sensible choice, as traditional methods are time-consuming, labor-intensive, and destructive. However, most predictive models utilizing remote sensing data are statistical rather than mechanistic, making them difficult to extend at interannual and regional scales. This study explored the interannual and regional transferability of the potato plant nitrogen content (PNC) prediction models, which combined environmental variables (EVs, e.g. temperature, precipitation, etc.) with proximal hyperspectral vegetation indices (VIs). Two methodologies were implemented to fuse EVs and VIs. The first involved a multiple regression analysis utilizing a multivariate linear model and a random forest approach, with VIs and EVs treated as independent variables, respectively. The second, a hierarchical linear model (HLM), employed EVs to dynamically adjust the relationship between VIs and PNC for different experimental sites. The predictive outcomes demonstrated that (i) the conventional method relying solely on optical VIs exhibited limited accuracy and stability in interannual and regional PNC forecasting; (ii) albeit the multivariate regression approach significantly enhanced model accuracy within the calibration set, its scalability across years and regions remained suboptimal; (iii) the HLM method exhibited high precision and scalability across years and regions, with R2, RMSE, and NRMSE values of 0.68, 0.50 %, and 19.68 % in the validation set, respectively. Those findings corroborate that using a two-tier HLM method can automatically adjust for discrepancies in VIs response to PNC through EVs, thereby enhancing the model's stability. Provided that remote sensing data and EVs are sustainably acquired over the potato growth cycle, it will provide a particularly promising approach to potato nitrogen diagnostics as a decision-making tool for regional application of nitrogen fertilizer.