Forecast Combination Approach Research Articles

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%.

Forecast combination approach with meta-fuzzy functions for forecasting the number of immigrants within the maritime line security project in Turkey.

In this study, forecasting the number of immigrants on the Turkey's maritime line for use in a national security project carried out by Turkish Government within the scope of fight against uncontrolled immigration is discussed for the first time. Handling with the immigration problem is one of the biggest concerns of Turkey as unsupervised immigration can adversely affect the demographic and economic structure of the country. Precautions are needed as the short-, medium- and long-term impacts of undetected immigrants on the country's ecosystem are unpredictable, but due to the uncertainties inherent in immigration, the cost of using government resources such as patrol vehicles to capture undocumented immigrants can be extremely high. In order to both minimize the expenditure problem and keep immigration under control by providing a proper scan, forecasting the number of immigrants on the maritime line route is seen as an important problem and studied by probabilistic and non-probabilistic models. Since the data for 2020 and 2021 could not be attained yet due to COVID-19, in order to obtain forecasts and compare actual observations for 2019, which is the primarily focus of the research in this study, the dataset of interest on the number of daily immigrants between years 2016 and 2019 is obtained from Turkish Coast Guard Command within Ministry of Interior of Republic of Turkey. To obtain the most accurate forecasts, seven distinguished forecasting methods, from simple to complex, are implemented. Then, the forecast combination approach with meta-fuzzy functions which combines all methods is proposed. Consequently, the forecasting results are acquired and evaluated by using R. The evaluation of the results is made by using widely considered measurement accuracy metric root mean square error. According to the final assessments, the proposed approach gives more accurate forecasting results for the expected number of immigrants on the Turkey's maritime line and these results become an input to the national security project.

An extreme bias-penalized forecast combination approach to commodity price forecasting

Forecast combination, a well-established technique for improving forecasting accuracy, investigates the integration of competing forecasts to produce a composite superior to individual forecasts. In this study, we propose a novel forecast combination method that would reduce overfitting risk and improve forecast’s generalization ability. To capture the extreme bias of a forecast in combination process, we define a measurement PaR for forecast combination. A novel PaR-based loss function with an elastic net is proposed that can effectively trade off the sparsity of weights to mitigate the risk of underfitting or overfitting. An improved artificial bee colony algorithm-based optimization method is introduced to achieve the optimal weights. The experimental results on gold, silver and crude oil price data demonstrate that the proposed forecast combination approach can outperform not only individual models but also combination approaches like simple averaging and other competitive benchmarks. The MAPE achieved by the presented method could decrease by 10.98%, 5.03% and 10.28% in gold, silver and crude oil price forecasting respectively, compared to the best individual model.

Multi-resolution, multi-horizon distributed solar PV power forecasting with forecast combinations

Distributed, small-scale solar photovoltaic (PV) systems are being installed at a rapidly increasing rate. This can cause major impacts on distribution networks and energy markets. As a result, there is a significant need for improved forecasting of the power generation of these systems at different time resolutions and horizons. However, the performance of forecasting models depends on the resolution and horizon. Forecast combinations (ensembles), that combine the forecasts of multiple models into a single forecast may be robust in such cases. Therefore, in this paper, we provide comparisons and insights into the performance of five state-of-the-art forecast models and existing forecast combinations at multiple resolutions and horizons. We propose a forecast combination approach based on particle swarm optimization (PSO) that will enable a forecaster to produce accurate forecasts for the task at hand by weighting the forecasts produced by individual models. Furthermore, we compare the performance of the proposed combination approach with existing forecast combination approaches. A comprehensive evaluation is conducted using a real-world residential PV power data set measured at 25 houses located in three locations in the United States. The results across four different resolutions and four different horizons show that the PSO-based forecast combination approach outperforms the use of any individual forecast model and other forecast combination counterparts, with an average Mean Absolute Scaled Error reduction by 3.81% compared to the best performing individual model. Our approach enables a solar forecaster to produce accurate forecasts for their application regardless of the forecast resolution or horizon.

Can machine learning improve small area population forecasts? A forecast combination approach

Generating accurate small area population forecasts is vital for governments and businesses as it provides better grounds for decision making and strategic planning of future demand for services and infrastructure. Small area population forecasting faces numerous challenges, including complex underlying demographic processes, data sparsity, and short time series due to changing geographic boundaries. In this paper, we propose a novel framework for small area forecasting which combines proven demographic forecasting methods, an exponential smoothing based algorithm, and a machine learning based forecasting technique. The proposed forecasting combination contains four base models commonly used in demographic forecasting, a univariate forecasting model specifically suitable for forecasting yearly data, and a globally trained Light Gradient Boosting Model (LGBM) that exploits the similarities between a collection of population time series. In this study, three forecast combination techniques are investigated to weight the forecasts generated by these base models. We empirically evaluate our method, by preparing small area population forecasts for Australia and New Zealand. The proposed framework is able to achieve competitive results in terms of forecasting accuracy. Moreover, we show that the inclusion of the LGBM model always improves the accuracy of combination models on both datasets, relative to combination models which only include the demographic models. In particular, the results indicate that the proposed combination framework decreases the prevalence of relatively poor forecasts, while improving the reliability of small area population forecasts.

Forecasting Tourism Demand With a New Time-Varying Forecast Averaging Approach

Existing research has shown that combination can effectively improve tourism forecasting accuracy compared with single model. However, the model uncertainty and structural instability in combination for out-of-sample tourism forecasting may influence the forecasting performance. This paper proposes a novel forecast combination approach based on time-varying jackknife model averaging (TVJMA), which can more efficiently handle structural changes and nonstationary trends in tourism data. Using Hong Kong tourism demand from five major tourism source regions as an empirical study, we investigate whether our proposed nonparametric TVJMA-based approach can improve tourism forecasting accuracy further. Empirical results show that the proposed TVJMA-based approach outperforms other competitors including single model and three combination methods in most cases. Findings indicate the outstanding performance of our method is robust to various forecasting horizons and different estimation periods.

Nowcasting Russian GDP using forecast combination approach

This paper presents a forecast combination approach for short-term assessment of economic growth in Russia. Our method significantly expands the existing domestic academic literature and embraces most advanced nowcasting techniques. The key feature of our approach is forecasting the growth rates of components GDP by expenditure using scenario indicators and a huge set of high-frequency predictors. The total number of combined models is nearly 500. Comparing our approach with popular benchmark models over the period from January 2003 to December 2020, we conclude that our method has the highest accuracy of short-term GDP forecasting. The findings of this research may be useful for the development of monetary and communication policy in Russia.

An adaptive forecast combination approach based on meta intuitionistic fuzzy functions

Intuitionistic meta fuzzy forecast combination functions are introduced in the paper. There are two challenges in the forecast combination literature, determining the optimum weights and the methods to combine. Although there are a few studies on determining the methods, there are numerous studies on determining the optimum weights of the forecasting methods. In this sense, the questions like “What methods should we choose in the combination?” and “What combination function or the weights should we choose for the methods” are handled in the proposed method. Thus, the first two contributions that the paper aims to propose are to obtain the optimum weights and the proper forecasting methods in combination functions by employing meta fuzzy functions (MFFs). MFFs are recently introduced for aggregating different methods on a specific topic. Although meta-analysis aims to combine the findings of different primary studies, MFFs aim to aggregate different methods based on their performances on a specific topic. Thus, forecasting is selected as the specific topic to propose a novel forecast combination approach inspired by MFFs in this study. Another contribution of the paper is to improve the performance of MFFs by employing intuitionistic fuzzy c-means. 14 meteorological datasets are used to evaluate the performance of the proposed method. Results showed that the proposed method can be a handy tool for dealing with forecasting problems. The outstanding performance of the proposed method is verified in terms of RMSE and MAPE.

Quantile regression averaging‐based probabilistic forecasting of daily ambient temperature

AbstractThe inclusion of conductor temperature variations for numerous power system planning and operational studies has long been recognized in the literature. The conductor temperature is majorly affected by environmental factors such as the ambient temperature. An efficient forecasting technique for forecasting ambient temperature is the need of the hour to prevent unexpected hazards in power systems and other areas caused due to temperature variations. Numerous researches have proposed different point forecasting and probabilistic forecasting models for the forecasting of ambient temperature. The probabilistic forecasting of ambient temperature provides complete information about future uncertainties, therefore quantifying the effects of daily temperature variations. A forecast combination approach, such as the quantile regression averaging, has never been utilized in this context. The selection of suitable point forecasters that complement each other's effects by characterizing different aspects of the ambient temperature data for averaging and the use of frequency components that explain the daily and periodic seasonal variations of ambient temperature to construct a forecasting model are the significant features of this paper. The proposed model is used with four varieties, and each is compared with the others. It is found that the variant using all the complementary individual point forecasters performs better in making probabilistic forecasts than the other options as well as better than the popular quantile k‐nearest neighbors, quantile egression forests, and basic quantile regression as inferred from the quantile score, Winkler score, and reliability plots.

Estimating portfolio risk for tail risk protection strategies

AbstractWe forecast portfolio risk for managing dynamic tail risk protection strategies, based on extreme value theory, expectile regression, copula‐GARCH and dynamic generalized autoregressive score models. Utilizing a loss function that overcomes the lack of elicitability for expected shortfall, we propose a novel expected shortfall (and value‐at‐risk) forecast combination approach, which dominates simple and sophisticated standalone models as well as a simple average combination approach in modeling the tail of the portfolio return distribution. While the associated dynamic risk targeting or portfolio insurance strategies provide effective downside protection, the latter strategies suffer less from inferior risk forecasts, given the defensive portfolio insurance mechanics.

Artificial bee colony-based combination approach to forecasting agricultural commodity prices

The fluctuation of agricultural commodity prices has attracted a considerable amount of attention. However, the complexity of the agricultural futures market and the variability of influencing factors makes the prediction of agricultural commodity futures prices difficult. We address the nonlinear characteristics of agricultural commodity futures price series by proposing a forecast combination approach based on a global optimization method, called the Artificial Bee Colony Algorithm (ABC), for forecasting soybean and corn futures prices. Firstly, we used three denoising techniques, namely singular spectral analysis (SSA), empirical mode decomposition (EMD), and variational mode decomposition (VMD), to filter the external noise in the original price series. Then, we generated diverse forecasting sub-models by combining denoising techniques and five popular forecasting models: autoregressive integrated moving average regression (ARIMA), support vector regression (SVR), recurrent neural network (RNN), gated recurrent neural network (GRU), and long-short term memory neural network (LSTM). Finally, we present an ABC approach for three forecast combinations: heterogeneous, semi-heterogeneous, and homogeneous combination. Experimental results indicate that the semi-heterogeneous forecast combination is superior to other combination strategies. For corn and soybean prices, ABC-based semi-heterogeneous forecast combinations have error reductions of 53.3% and 50.0% of MAPE and improvements of 32.4% and 34.5% in Dstat compared to the best single models, respectively.

Exploring an Ensemble of Methods that Combines Fuzzy Cognitive Maps and Neural Networks in Solving the Time Series Prediction Problem of Gas Consumption in Greece

This paper introduced a new ensemble learning approach, based on evolutionary fuzzy cognitive maps (FCMs), artificial neural networks (ANNs), and their hybrid structure (FCM-ANN), for time series prediction. The main aim of time series forecasting is to obtain reasonably accurate forecasts of future data from analyzing records of data. In the paper, we proposed an ensemble-based forecast combination methodology as an alternative approach to forecasting methods for time series prediction. The ensemble learning technique combines various learning algorithms, including SOGA (structure optimization genetic algorithm)-based FCMs, RCGA (real coded genetic algorithm)-based FCMs, efficient and adaptive ANNs architectures, and a hybrid structure of FCM-ANN, recently proposed for time series forecasting. All ensemble algorithms execute according to the one-step prediction regime. The particular forecast combination approach was specifically selected due to the advanced features of each ensemble component, where the findings of this work evinced the effectiveness of this approach, in terms of prediction accuracy, when compared against other well-known, independent forecasting approaches, such as ANNs or FCMs, and the long short-term memory (LSTM) algorithm as well. The suggested ensemble learning approach was applied to three distribution points that compose the natural gas grid of a Greek region. For the evaluation of the proposed approach, a real-time series dataset for natural gas prediction was used. We also provided a detailed discussion on the performance of the individual predictors, the ensemble predictors, and their combination through two well-known ensemble methods (the average and the error-based) that are characterized in the literature as particularly accurate and effective. The prediction results showed the efficacy of the proposed ensemble learning approach, and the comparative analysis demonstrated enough evidence that the approach could be used effectively to conduct forecasting based on multivariate time series.

Out-of-sample equity premium prediction: a complete subset quantile regression approach

This paper extends the complete subset linear regression framework to a quantile regression setting. We employ complete subset combinations of quantile forecasts in order to construct robust and accurate equity premium predictions. We show that our approach delivers statistically and economically significant out-of-sample forecasts relative to both the historical average benchmark, the complete subset mean regression approach and the single-variable quantile forecast combination approach. Our recursive algorithm that selects, in real time, the best complete subset for each predictive regression quantile succeeds in identifying the best subset in a time- and quantile-varying manner.

Merging radar and gauge information within a dynamical model combination framework for precipitation estimation in cold climates

This study presents a dynamic forecast combination approach adapted to incorporate multiple sources of precipitation. Dynamic combination serves to utilise the varying merit each data source exhibits with time. The dynamic model combination framework presented merges a nonparametric k-nearest neighbour (k-nn) estimation of radar precipitation with Thin Plate Spline (TPS) interpolated gauge precipitation. Since air temperature is an essential variable to discriminate the phase of the precipitation in cold climates, this study uses radar precipitation and air temperature as the two variables in the dynamic combination algorithm. The merging of k-nn and TPS estimates is shown to reduce the RMSE by 25% compared to the original radar precipitation rates. The usefulness of air temperature is found not to be as significant in the combination as it is in the formulation of the nonparametric radar precipitation fields for cold incident temperatures.

Are advanced emerging market stock returns predictable? A regime-switching forecast combination approach

Advanced emerging markets (AEMs) transitioning into developed markets experience far-reaching economic and institutional changes. Developing predictive models of stock returns in AEMs involves challenges of parameter instability and model uncertainty. This study uses Markov regime switching (MRS) models to address parameter instability and a combination forecast approach to mitigate model uncertainty. We find that the MRS model better captures the effects of predictor variables on returns compared to models with time-invariant parameters and produces statistically and economically significant return forecasts. Combining return forecasts from different MRS models further improves return predictability in AEMs. Consequently, employing MRS models in conjunction with the combination forecast approach goes a long way to improving forecast accuracy in AEMs.

Краткосрочное прогнозирование поступлений в бюджет с использованием комбинации прогнозов

In this paper the author compares a number of methods to forecast corporate income tax revenues in the next quarter: autoregressive integrated moving average, exponential smoothing, linear regression, naive forecast, and combination approaches. Results show that the increase of the time period for the next quarter forecast error calculation leads to the increase of errors for individual models and arithmetic mean, and the decrease for combination approaches which consider forecast errors in previous periods. Linear regression with the MOEX Russia Index as explanatory variable provides the lowest error for the next quarter forecast for 5- and 10-quarter periods. The forecast combination approach, what takes into account the forecast error in previous 4 quarters, provides the best result for a period of 15 quarters, what is explained by diversification of forecast errors. This method can be successfully used for corporate income tax revenue projections and possibly for other budget revenues.

Much Ado about the Egyptian Pound: Exchange Rate Misalignment and the Path Towards Equilibrium

Abstract This paper estimates Egypt's equilibrium real exchange rate and exchange rate misalignment based on economic fundamentals over the period 2001Q3–2017Q3. Focusing on the more recent period, we find that the Egyptian pound was undervalued by about 22.3% in 2017Q1 due to overshooting its equilibrium value after floating the currency in 2016Q4. The currency undervaluation then declined to 18.5% in 2017Q3 driven by an increase in the real effective exchange rate due to a surge in domestic inflation. With regard to the determinants of the equilibrium real exchange rate, we find the productivity differential (vis-à-vis Egypt's trade partners) and trade openness to be the most significant factors. We also provide projections for the equilibrium real exchange rate and exchange rate misalignment until 2020Q4, which reveal that the exchange rate undervaluation will be dissipating fast due to high inflation. If the nominal exchange rate stabilizes at its level in 2017Q3 (17.73 pounds per US dollar), the currency will be overvalued by 13.1% in 2020Q4. Given the uncertainty surrounding the projections, a forecast combination approach is also presented. Finally, the paper highlights the implications of the empirical findings for the conduct of monetary policy in Egypt.

Combining predictive distributions for the statistical post-processing of ensemble forecasts

Statistical post-processing techniques are now used widely for correcting systematic biases and errors in the calibration of ensemble forecasts obtained from multiple runs of numerical weather prediction models. A standard approach is the ensemble model output statistics (EMOS) method, which results in a predictive distribution that is given by a single parametric law, with parameters that depend on the ensemble members. This article assesses the merits of combining multiple EMOS models based on different parametric families. In four case studies with wind speed and precipitation forecasts from two ensemble prediction systems, we investigate the performances of state of the art forecast combination methods and propose a computationally efficient approach for determining linear pool combination weights. We study the performance of forecast combination compared to that of the theoretically superior but cumbersome estimation of a full mixture model, and assess which degree of flexibility of the forecast combination approach yields the best practical results for post-processing applications.

Estimating Portfolio Risk for Tail Risk Protection Strategies

We forecast portfolio risk for managing dynamic tail risk protection strategies, based on extreme value theory, expectile regression, Copula-GARCH and dynamic GAS models. Utilizing a loss function that overcomes the lack of elicitability for Expected Shortfall, we propose a novel Expected Shortfall (and Value-at-Risk) forecast combination approach, which dominates simple and sophisticated standalone models as well as a simple average combination approach in modelling the tail of the portfolio return distribution. While the associated dynamic risk targeting or portfolio insurance strategies provide effective downside protection, the latter strategies suffer less from inferior risk forecasts given the defensive portfolio insurance mechanics.

Measuring the market risk of freight rates: A forecast combination approach

AbstractThis paper addresses the issue of freight rate risk measurement via value at risk (VaR) and forecast combination methodologies while focusing on detailed performance evaluation. We contribute to the literature in three ways: First, we reevaluate the performance of popular VaR estimation methods on freight rates amid the adverse economic consequences of the recent financial and sovereign debt crisis. Second, we provide a detailed and extensive backtesting and evaluation methodology. Last, we propose a forecast combination approach for estimating VaR. Our findings suggest that our combination methods produce more accurate estimates for all the sectors under scrutiny, while in some cases they may be viewed as conservative since they tend to overestimate nominal VaR.