Individual Forecasting Models Research Articles

CONCEPT OF AN ENSEMBLE FORECASTING SYSTEM FOR OPTIMIZATION PROBLEMS OF CONTROL OF SOLAR MICROGRID

Accurate probabilistic forecasts of renewable generation are the driving force for optimizing the operation and management of MicroGrid systems. Combining forecasts of different individual models can improve forecast accuracy, but unlike combining point forecasts, for which simple weighted averaging is often a plausible solution, combining probabilistic forecasts is a much more complex task. Today, ensembles of forecasting models are one of the promising directions for problem solving, where forecasting accuracy is more important than the ability to interpret the model. The main idea of ensembles is the training of several basic models and the aggregation of the results of their work. Empirical studies show that combinations of forecasts, on average, are more likely to produce better forecasts than methods that are based on selecting only one forecasting model. When building ensembles, the issue of ensuring diversity of models and effective training of model members of the ensemble becomes especially relevant. The article is devoted to solving the issues of building an ensemble model for forecasting photovoltaic (PV) power, which combines the results of several basic probabilistic models. Using the ensemble method proposed by the authors can improve forecasting accuracy and reduce the time required for training and evaluation of ensemble member models. Directions and prospects of further research are formulated.

An attention-PCA based forecast combination approach to crude oil price

Crude oil price forecasting has garnered considerable attention due to its pivotal role in both market dynamics and economic stability. In this study, we present an attention-based principal component analysis (attention-PCA) methodology designed to improve the performance of oil price forecasting models. The attention-PCA approach enables greater focus on predictor variables with superior forecasting capabilities. Furthermore, we develop a diversity enhancement mechanism for forecast combination by incorporating multiple attention mechanisms, varying numbers of principal components, and a range of forecasting models. The empirical results demonstrates that attention-PCA-based individual forecasting models significantly outperform benchmark models, reducing the Mean Absolute Percentage Error (MAPE) by up to 43.2%. The proposed forecast combination strategy yields the most accurate and diverse forecasts among those evaluated, with the MAPE of the optimal combination model standing at 4.40%.

Forecasting Day-Ahead Electricity Prices for the Italian Electricity Market Using a New Decomposition—Combination Technique

Over the last 30 years, day-ahead electricity price forecasts have been critical to public and private decision-making. This importance has increased since the global wave of deregulation and liberalization in the energy sector at the end of the 1990s. Given these facts, this work presents a new decomposition–combination technique that employs several nonparametric regression methods and various time-series models to enhance the accuracy and efficiency of day-ahead electricity price forecasting. For this purpose, first, the time-series of the original electricity prices deals with the treatment of extreme values. Second, the filtered series of the electricity prices is decomposed into three new subseries, namely the long-term trend, a seasonal series, and a residual series, using two new proposed decomposition methods. Third, we forecast each subseries using different univariate and multivariate time-series models and all possible combinations. Finally, the individual forecasting models are combined directly to obtain the final one-day-ahead price forecast. The proposed decomposition–combination forecasting technique is applied to hourly spot electricity prices from the Italian electricity-market data from 1 January 2014 to 31 December 2019. Hence, four different accuracy mean errors—mean absolute error, mean squared absolute percent error, root mean squared error, and mean absolute percent error; a statistical test, the Diebold–Marino test; and graphical analysis—are determined to check the performance of the proposed decomposition–combination forecasting method. The experimental findings (mean errors, statistical test, and graphical analysis) show that the proposed forecasting method is effective and accurate in day-ahead electricity price forecasting. Additionally, our forecasting outcomes are comparable to those described in the literature and are regarded as standard benchmark models. Finally, the authors recommended that the proposed decomposition–combination forecasting technique in this research work be applied to other complicated energy market forecasting challenges.

A modified weighting system for combined forecasting methods based on the correlation coefficients of the individual forecasting models

Herein, a modified weighting for combined forecasting methods is established. These weights are used to adjust the correlation coefficient between the actual and predicted values from five individual forecasting models based on their correlation coefficient values and ranking. Time-series datasets with three patterns (stationary, trend, or both trend and seasonal) were analyzed by using the five individual forecasting models and three combined forecasting methods: simple-average, Bates-Granger, and the proposed approach. The MAPE and RMSE results indicate that the proposed method outperformed the others, especially when the time-series pattern was stationary and improved the forecasting accuracy of the worst and best individual forecasting models by 35–37% and 7–10%, respectively. Moreover, the proposed method showed improvements in MAPE and RMSE of around 18–20% and 9–11% compared to the simple-average and Bates-Granger methods, respectively. In addition, the combined forecasting methods outperformed the individual forecasting models when analyzing non-stationary data. Remarkably, the performances of the proposed and Bates-Granger methods were almost the same, with improvements in MAPE and RMSE in the range of 1–2% on average. Therefore, the proposed method for creating weights based on the correlation coefficients of the individual forecasting models greatly improves combined forecasting methods.

Forecasting the PV Power Utilizing a Combined Convolutional Neural Network and Long Short-Term Memory Model

The quantity of photovoltaic systems generated relies on the climate, and to generate significant variations depending on the climatic conditions. Its electricity production processes exhibit unpredictability, instability, and intermittent behavior. The augmented convolutional long-short memory combination network is recommended for PV energy combined forecasting in accordance to overcome the challenge of considerable instability in the high proportion of sustainable energy output. The LSTM and CNN models’ training procedure uses the enhanced artificial bee colony (ABC) to adjust their characteristics. To determine the projected model parameters of PV power, the estimated input parameters of every model are stacked and then reassembled. The studies showed that the forecast precision was significantly higher than the reliability rate of the Convolutional LSTM combinational network as well as the forecasting findings from the BP, CNN, and LSTM independent computational methods. This forecast specificity was predicated on the enhanced ABC to optimize the variables of the CLSTM combinational network for forecasting Photo Voltaic power. For comparative study and assessment, the most recent operational data from a Photovoltaic power plant in northern China have been utilized. The suggested technique resolved more quickly and was higher weather-adaptable than the individual neural network and hybrid forecasting models mentioned prior.

A dynamic clustering ensemble learning approach for crude oil price forecasting

Accurate oil price forecasts matter, yet the nonstationarity of oil prices makes forecasting a challenging task. In this study, we propose a dynamic ensemble forecasting method for nonstationary oil prices using clustering approaches. Specifically, clustering is embedded in the ensemble forecasting framework, whereby the given period of historical observations is automatically classified into several clusters according to the data characteristics. This classification provides a solid groundwork for dynamically evaluating individual forecasting models in a targeted manner. We then propose a clustering-based regular increasing monotone weight assignment strategy that removes the influence of outliers and assigns appropriate weights to each forecasting model, thereby balancing the competitiveness and robustness of the proposed ensemble model. We verify the competitiveness and robustness of the proposed model by using West TX Intermediate oil prices. Results show that the proposed model significantly outperforms benchmarks and state-of-the-art methods in terms of horizontal and directional accuracy and is thus competitive. The robustness of the proposed model is validated using scenarios involving parameter variation and data missing assumptions. In summary, we present a model with promising effectiveness in promoting prediction performance in forecasting oil prices.

Forecasting stock volatility with a large set of predictors: A new forecast combination method

AbstractThis paper investigates how to improve the prediction accuracy of stock realized volatility using a large set of predictors. Exploiting normalized positive adjusted and significant statistic of predictor obtained from the in‐sample result as weight, we develop two simple and effective forecast combination methods. Using an array of 86 equity, bond, forex, futures, behavior, macroeconomic, and uncertainty predictors, we find that the proposed methodologies significantly improve stock realized volatility out‐of‐sample prediction performance relative to several extant forecast combinations. This result is robust for different individual forecast models, different dependent variables, and different out‐of‐sample periods. Furthermore, we explain that out‐of‐sample predictability varies significantly with changes in the number of predictors. And the existence of a strongly powerful volatility predictor affects this change.

Two-Stage Portfolio Optimization Integrating Optimal Sharp Ratio Measure and Ensemble Learning

The traditional portfolio theory has relied heavily on historical asset returns while ignoring future information. Based on ensemble learning and maximum Sharpe ratio portfolio theory, this paper proposes a two-stage portfolio optimization method by considering asset forecast information, aiming to improve the performance and robustness of a portfolio. In the first stage, concerning the underlying asset selection, we integrate six individual prediction models using the ensemble learning method to forecast the future return of assets where the assets with higher potential returns are selected for portfolio optimization. In the second stage, we propose a novel investment strategy by combining the forecasted returns of selected assets with the maximum Sharpe ratio portfolio model. In the empirical analysis, we employ the constituent stocks of the China Securities Index 300 (CSI 300) to test the out-of-sample performance of the proposed strategy with several traditional portfolio strategies, including minimum variance portfolio strategy, traditional maximum Sharpe ratio portfolio strategy, 1/N portfolio strategy, and CSI 300 index. Our analytical results show that (1) compared to individual forecasting models, the ensemble learning method is more accurate in forecasting stock returns, and (2) the proposed portfolio strategy largely outperforms most of its competitors in terms of the Sharpe ratio, Sortino ratio, Omega ratio, and Calmar ratio. This indicates that the proposed two-stage portfolio optimization method is of potential to construct a promising investment strategy due to its trade-off between historical and future information of assets.

Forecasting VaR and ES using the joint regression combined forecasting model in the Chinese stock market

PurposeThis paper aims to better jointly estimate Value at Risk (VaR) and expected shortfall (ES) by using the joint regression combined forecasting (JRCF) model.Design/methodology/approachCombining different forecasting models in financial risk measurement can improve their prediction accuracy by integrating the individual models’ information. This paper applies the JRCF model to measure VaR and ES at 5%, 2.5% and 1% probability levels in the Chinese stock market. While ES is not elicitable on its own, the joint elicitability property of VaR and ES is established by the joint consistent scoring functions, which further refines the ES’s backtest. In addition, a variety of backtesting and evaluation methods are used to analyze and compare the alternative risk measurement models.FindingsThe empirical results show that the JRCF model outperforms the competing models. Based on the evaluation results of the joint scoring functions, the proposed model obtains the minimum scoring function value compared to the individual forecasting models and the average combined forecasting model overall. Moreover, Murphy diagrams’ results further reveal that this model has consistent comparative advantages among all considered models.Originality/valueThe JRCF model of risk measures is proposed, and the application of the joint scoring functions of VaR and ES is expanded. Additionally, this paper comprehensively backtests and evaluates the competing risk models and examines the characteristics of Chinese financial market risks.

A Study on House Price Prediction Based on Stacking-Sorted-Weighted-Ensemble Model

<p>The house price prediction problem is a typical regression problem, and most of the common house price prediction models are single prediction algorithms, which are not ideal in terms of accuracy and stability. For solving this problem, this paper proposes a house price forecasting method based on Stacking-Sorted-Weighted-Ensemble (SSWE) model. Considering the characteristics of different algorithms and giving full play to the advantages of each model, multiple individual forecasting models are fused with the Stacking model. The algorithm validation is performed using the data generated by the system of real estate management department in western Guangdong. The prediction results show that the Stacking model is superior to the single model. Compared with the Stacking regression model, the SSWE model has a 13.6% increase in the root mean square error on the training set but a 0.3% decrease on the test set, indicating that the SSWE model prevents overfitting to a some extent and increases the accuracy and stability of the model.</p> <p> </p>

An Intelligent Time-Series Model for Forecasting Bus Passengers Based on Smartcard Data

Public transportation systems are an effective way to reduce traffic congestion, air pollution, and energy consumption. Today, smartcard technology is used to shorten the time spent boarding/exiting buses and other types of public transportation; however, this does not alleviate all traffic congestion problems. Accurate forecasting of passenger flow can prevent serious bus congestion and improve the service quality of the transportation system. To the best of the current authors’ knowledge, fewer studies have used smartcard data to forecast bus passenger flow than on other types of public transportation, and few studies have used time-series lag periods as forecast variables. Therefore, this study used smartcard data from the bus system to identify important variables that affect passenger flow. These data were combined with other influential variables to establish an integrated-weight time-series forecast model. For different time data, we applied four intelligent forecast methods and different lag periods to analyze the forecasting ability of different daily data series. To enhance the forecast ability, we used the forecast data from the top three of the 80 combined forecast models and adapted their weights to improve the forecast results. After experiments and comparisons, the results show that the proposed model can improve passenger flow forecasting based on three bus routes with three different series of time data in terms of root-mean-square error (RMSE) and mean absolute percentage error (MAPE). In addition, the lag period was found to significantly affect the forecast results, and our results show that the proposed model is more effective than other individual intelligent forecast models.

A Study on Load Forecasting of Distribution Line Based on Ensemble Learning for Mid- to Long-Term Distribution Planning

The complexity and uncertainty of the distribution system are increasing as the connection of distributed power sources using solar or wind energy is rapidly increasing, and digital loads are expanding. As these complexity and uncertainty keep increasing the investment cost for distribution facilities, optimal distribution planning becomes a matter of greater focus. This paper analyzed the existing mid-to-long-term load forecasting method for KEPCO’s distribution planning and proposed a mid- to long-term load forecasting method based on ensemble learning. After selecting optimal input variables required for the load forecasting model through correlation analysis, individual forecasting models were selected, which enabled the derivation of the optimal combination of ensemble load forecast models. This paper additionally offered an improved load forecasting model that considers the characteristics of each distribution line for enhancing the mid- to long-term distribution line load forecasting process for distribution planning. The study verified the performance of the proposed method by comparing forecasting values with actual values.

Quantifying Time-Varying Forecast Uncertainty and Risk for the Real Price of Oil

We propose a novel and numerically efficient quantification approach to forecast uncertainty of the real price of oil using a combination of probabilistic individual model forecasts. Our combination method extends earlier approaches that have been applied to oil price forecasting, by allowing for sequentially updating of time-varying combination weights, estimation of time-varying forecast biases and facets of miscalibration of individual forecast densities and time-varying inter-dependencies among models. To illustrate the usefulness of the method, we present an extensive set of empirical results about time-varying forecast uncertainty and risk for the real price of oil over the period 1974–2018. We show that the combination approach systematically outperforms commonly used benchmark models and combination approaches, both in terms of point and density forecasts. The dynamic patterns of the estimated individual model weights are highly time-varying, reflecting a large time variation in the relative performance of the various individual models. The combination approach has built-in diagnostic information measures about forecast inaccuracy and/or model set incompleteness, which provide clear signals of model incompleteness during three crisis periods. To highlight that our approach also can be useful for policy analysis, we present a basic analysis of profit-loss and hedging against price risk.

Smoothed Bernstein Online Aggregation for Short-Term Load Forecasting in IEEE DataPort Competition on Day-Ahead Electricity Demand Forecasting: Post-COVID Paradigm

We present a winning method of the IEEE DataPort Competition on Day-Ahead Electricity Demand Forecasting: Post-COVID Paradigm. The day-ahead load forecasting approach is based on a novel online forecast combination of multiple point prediction models. It contains four steps: i) data cleaning and preprocessing, ii) a new holiday adjustment procedure, iii) training of individual forecasting models, iv) forecast combination by smoothed Bernstein Online Aggregation (BOA). The approach is flexible and can quickly adjust to new energy system situations as they occurred during and after COVID-19 shutdowns. The ensemble of individual prediction models ranges from simple time series models to sophisticated models like generalized additive models (GAMs) and high-dimensional linear models estimated by lasso. They incorporate autoregressive, calendar, and weather effects efficiently. All steps contain novel concepts that contribute to the excellent forecasting performance of the proposed method. It is especially true for the holiday adjustment procedure and the fully adaptive smoothed BOA approach.

Forecasting crop yields through climate variables using mixed frequency data. The case of Argentine soybeans

This article evaluates the value of information on climate variables published in advance and at a higher frequency than the target variable of interest -crop yields- in order to get short-term forecasts. Aggregate and disaggregate climate data, alternative weighting schemes and di erent updating schemes are used to evaluate forecasting performance. This study focuses on the case of soybean yields in Argentina. Results show that models including high frequency weather data outperformed particularly during the three consecutive compaigns after 2008/09 when soybean yield decreased almost by 50%. Furthermore, forecast combinations showed a better forecasting performance than individual forecasting models.

A multi-model data fusion methodology for seasonal drought forecasting under uncertainty: Application of Bayesian maximum entropy

In this paper, we present a new methodology for improving the results of seasonal drought forecasting by developing a Bayesian Maximum Entropy-based fusion (BMEF) model. The BMEF model combines the forecasts done by four individual (single-source) data-driven models to achieve better outcomes. Regional drought indices of Effective Drought Index (EDI) and Multiple Standard Precipitation Index (MSPI) are forecasted using the individual forecasting models of Artificial Neural Network (ANN), Adaptive Neuro Fuzzy Inference System (ANFIS), Support Vector Regression (SVR), and M5tree. The outputs of the individual models with the best performances are selected to be fused using the BMEF model and the results are analyzed and compared. The effect of different large-scale climate signals on rainfall and drought forecasting is analyzed and the most effective climate variables are selected as predictors in the forecasting models. Next, the uncertainty analysis on the results of the individual models as well as those of the BMEF model is carried out by deriving the probability mass functions of the drought indices using a resampling technique and Monte Carlo analysis. Finally, the results of the uncertainty analysis are evaluated to compare the performance of individual models and the BME-based fusion model in decreasing the uncertainty of seasonal drought forecasting. The performance of the proposed methodology is evaluated by using it to forecast seasonal drought conditions in the southwest of Iran. Based on the results of the uncertainty analysis, the BMEF model provides more reliable forecasts particularly for severe drought events than the individual models. It is also inferred that adding the SST to the predictors, decreases the uncertainty of drought forecasts.

Forecasting stock market volatility using commodity futures volatility information

By incorporating volatility information from nineteen commodity futures prices, this paper compares the predictive ability of traditional individual AR-type and combination forecasting models versus model shrinkage methods in predicting US stock market volatility. Our empirical results show that the Lasso shrinkage method has significantly better out-of-sample forecasting performance in not only the individual models but also the combination approaches. In particular, the Lasso model with all predictors exhibits the best out-of-sample forecasting performance, suggesting that incorporating all commodity futures volatility information by the model shrinkage approach is an effective way for market participants and policy-makers to obtain accurate forecasts of US stock market volatility. Further analysis shows that the predictability evidence is substantially clearer during high volatility periods than in low volatility regimes. Finally, alternative evaluation periods further confirm the robustness of our results.

Novel Cooperative Multi-Input Multilayer Perceptron Neural Network Performance Analysis with Application of Solar Irradiance Forecasting

To forecast solar irradiance with higher accuracy and generalization capability is challenging in the photovoltaic (PV) energy system. Meteorological parameters are highly influential in solar irradiance, leading to intermittent and randomicity. Forecasting using a single neural network model does not have sufficient generalization ability to achieve the optimal forecasting of solar irradiance. This paper proposes a novel cooperative multi-input multilayer perceptron neural network (CMMLPNN) to mitigate the issues related to generalization and meteorological effects. Authors develop a proposed forecasting neural network model based on the amalgamation of two inputs, three inputs, four inputs, five inputs, and six inputs associated multilayer perceptron neural network. In the proposed forecasting model (CMMLPNN), the authors overcome the variance based on the meteorological parameters. The amalgamation of five multi-input multilayer perceptron neural networks leads to better generalization ability. Some individual multilayer perceptron neural network-based forecasting models outperform in some situations, but cannot assure generalization ability and suffer from the meteorological weather condition. The proposed CMMLPNN (cooperative multi-input multilayer perceptron neural network) achieves better forecasting accuracy with the generalization ability. Therefore, the proposed forecasting model is superior to other neural network-based forecasting models and existing models.

Solar and wind power generation forecasts using elastic net in time-varying forecast combinations

Precise renewable energy feed-in forecasts are essential for an effective and efficient integration of renewables into energy systems, and research contributions that help to reduce the uncertainty related to renewables are in high demand. This importance will increase in the future, as renewable energies are the world’s fastest growing electricity generation capacities. Forecast combinations have been empirically proven to outperform individual forecasting models in many disciplines. Our work uses an elastic net method, with cross-validation and rolling window estimation, in the context of renewable energy forecasts. Namely, the forecast combinations are obtained using regional data from Germany for both solar photovoltaic and wind feed-in during the period 2010–2018, with quarter-hourly frequency. The dynamic elastic net estimation, preceded by dynamic data pre-processing, improves forecasting accuracy for both photovoltaic and wind power feed-in forecasts. Moreover, our forecasting framework outperforms benchmarks such as simple average and individual forecasts. Our forecasting framework can be applied widely to estimate renewable power in other countries, systems, or individual power plants.

Short-term nodal voltage forecasting for power distribution grids: An ensemble learning approach

The integration of distributed energy resources (DER) complicates the operation of the power distribution grids, and the nodal voltage may violate frequently. Making accurate predictions of the nodal voltage is fundamental for voltage regulation of the distribution grid. Even though energy forecasting has been widely studied, voltage is still a rarely touched area. This paper enriches the research by proposing an ensemble approach for both deterministic and probabilistic short-term nodal voltage forecasting. Specifically, a new joint model- and data-driven feature selection is first performed to select the most relevant features for distribution grid voltage forecasting. Then, different individual forecasting models are trained using the selected features. On this basis, simple weighted averaging and quantile regression averaging approaches are applied to combine the individual models for deterministic and probabilistic forecasting, respectively. Finally, case studies are conducted on a real-world distribution grid to verify the effectiveness and superiority of the proposed method.