Improvement In Forecast Accuracy Research Articles

An empirical investigation of comparative performance of approximate and exact corrections of the bias in Croston's method in forecasting lumpy demand

A positive bias in Croston's method, which had been developed to forecast intermittent demand, was reported by Syntetos and Boylan (2001). They proposed an approximate correction. Subsequently, Shale et al. (2006) proposed an 'exact' correction. Both corrections were derived analytically. The mathematical analysis establishes the superiority of the exact correction over both Croston's method and the approximate correction. We empirically investigate whether or not there are significant improvements in statistical forecast accuracy as well as inventory control performance obtained by applying the approximate or exact correction when forecasting lumpy demand. Using extensive simulation experiments, we find overall superior forecast accuracy of the bias correction methods over both simple exponential smoothing and Croston's methods. However, the exact correction yielded the same or only marginally better accuracy measures compared with the approximate correction. Moreover, in terms of inventory control performance, we observe marginal differences in inventory on hand and backlogs.

Factors that affect the improvement of demand forecast accuracy through point-of-sale reporting

Recent research has examined the use of real-time, shared point-of-sale (POS) data in forecasting. Although initial research posited that POS data can improve forecast accuracy, recent research has called some of these findings into question. We identify item order quantity, item order variability, and item order frequency of orders as specific factors that can affect the degree of improvement in POS demand forecasting accuracy when compared to order history-based techniques. Using a hierarchical linear model, we examine 60,651 orders for hundreds of items across 25 different distribution centers. We find a 11.2% overall improvement in using real-time, shared POS data in demand forecasting over order history-based forecasting. However, we find a curvilinear relationship between these improvements and both the order quantity and the item order variability. Additionally, we find that POS based forecasting improvements are greatest (1) when items are not frequently ordered, (2) when there is low variance in the number of distribution center ordering an item each week, and (3) when order quantities are neither relatively high, nor relatively low.

Earnings Announcement Clustering and Analyst Forecast Behavior

We examine how a limited attention effect can directly affect the information processing of a sophisticated market participant. Specifically, we investigate whether analysts respond to earnings announcement events differently if more than one firm in their coverage portfolios announce earnings on the same day (“Busy Analysts”). We find evidence that, for the same firm, analysts delay or even skip issuing earnings forecasts in the quarters when they are busy versus quarters when they are not. Compared with “Not-busy Analysts”, the initial responses of Busy Analysts are of lower forecast quality (in terms of forecast accuracy improvements, forecast boldness, and the price impact of forecasts). Further, we find that when analysts face multiple concurrent earnings announcements, they tend to prioritize issuing forecasts for firms with richer information environments over firms with weaker information environments. At the firm level, when a larger percentage of analysts covering the firm are busy, the analysts as a whole are less responsive to the firm’s earnings announcements, implying that the price discovery after the information event is slower. Our study extends the literatures of limited attention and analysts’ information production.

Projections of Inflation Dynamics for Pakistan: GMDH Approach

Abstract. This study is focused on identifying, based on various forecast accuracy criteria, best inflation forecasting model for Pakistan using the in sample projections for Pakistan inflation from 2006II to 2009II. To resolve the important issue of degree of contribution in forecasting performance of the two monetary aggregates in forecasting inflation, three main predictors: real GDP, interest rate and one out of the two monetary aggregate have been used, thus constructing two models; one with Divisia Monetary Index (DMI) and other with Simple sum monetary aggregate (SSMA). It is revealed that, though both of the monetary aggregates are important predictors in forecasting inflation, but DMAs provide better fit and improved forecasts as compared to their simple sum counterpart. Hence, the evidence is established that monetary aggregates still play a dominant role in predicting inflation for Pakistan economy. The study recommends the construction, publication, and use of high frequency DMAs by the State Bank of Pakistan (SBP) for forecasting inflation in Pakistan instead of SSMAs. Finally, to identify the improvement in forecast accuracy w.r.t. different forecasts combination, these forecasts have been combined and compared. It is revealed that when the structure of an empirically observed underlying series has complex nonlinear structure then forecasts based on single nonlinear model may fail to capture these diverse complexities. The best strategy is then to use various nonlinear models and combine these forecasts. Further the study concluded that if the complex nonlinear structure of an observed series is, a priory, unknown then universal approximators like Group Method of Data Handling (GMDH)- Polynomial Neural Networks (PNNs) and GMDH-Combinatorially Optimized (CO) could provide outstandingly accurate forecasts yet avoiding ‘overfitting’ even for small sample size. Specifically, it recommends the use of nonlinear non-parametric universal approximators for forecasting inflation in Pakistan by the SBP. Keywords. Monetary aggregate, Nonparametric nonlinear models, Universal approximators, Forecasting performance, Forecasts combination. JEL. E31, E47, E51, E52.

A Contextual Evaluation of Composite Forecasts of Annual Earnings

This study addresses the impact of firm- and time-specific attributes on the accuracy of composite forecasts of annual earnings, constructed from time-series, price-based, and analysts' forecasts. The attributes examined include firm size, analysts' coverage, and time periods pre-dating and following the implementation of regulation fair disclosure. Our results indicate that the relative accuracy of the composite forecasts is time-specific. In the pre-regulation fair disclosure period, composite forecasts significantly outperform each of the three individual forecast sources. Moreover, the extent of improvement in accuracy of composite forecasts is significantly higher for the smaller and lightly-covered firms. Collectively, these results suggest that the predictive accuracy of composite forecasts is contextual.

On the Accuracy of the Private Sector Macroeconomic Forecasts of the South African Economy

This paper evaluates the forecasting accuracy of private sector forecasters who participate in the annual “Media24 Economist of the Year” forecasting competition in South Africa. Our primary aim is to examine whether the accuracy of private sector forecasters improved over time, particularly their ability to predict the 2008/2009 recession and whether there was a distinct change in forecasting accuracy along this turning point of the business cycle. Our estimates from a forecasting error measure known as Theil's U-Statistic show that, on the average, the Root Weighted Mean Squared Error (RWMSE) of the growth forecast for the current period was 0.62 of an adaptive-naïve forecast, whereas the inflation forecast was 0.71 of an adaptive-naïve forecast. In order to determine whether there was an improvement in forecasting accuracy after the recession, we segregate the sample period along this break date and compare the size of the forecast errors between the two periods. To this end, we find that with respect to the growth predictions, there was a marginal reduction in the magnitude of the forecasting errors. However, in the case of inflation forecasts (both current and year-ahead), there was a marked reduction in the size of both the RWMSE and Theil's U-Statistic, implying that the post recessionary period was characterized by an improvement in accuracy of inflation forecasts made by the private sector. Furthermore, with respect to year-ahead forecasts, the results from both the nonparametric and the more formal parametric test allows us to reject the hypothesis of equality between the mean squared errors of the competing forecasts. Although the private sector forecasters were unable to accurately predict the recession, they were at the least able to produce forecasts that were more accurate than the adaptive-naïve model.

특수일 조업률 반영을 통한 전력수요예측 정확도 향상

특수일 조업률 반영을 통한 전력수요예측 정확도 향상

Hybridization of Chaotic Quantum Particle Swarm Optimization with SVR in Electric Demand Forecasting

In existing forecasting research papers support vector regression with chaotic mapping function and evolutionary algorithms have shown their advantages in terms of forecasting accuracy improvement. However, for classical particle swarm optimization (PSO) algorithms, trapping in local optima results in an earlier standstill of the particles and lost activities, thus, its core drawback is that eventually it produces low forecasting accuracy. To continue exploring possible improvements of the PSO algorithm, such as expanding the search space, this paper applies quantum mechanics to empower each particle to possess quantum behavior, to enlarge its search space, then, to improve the forecasting accuracy. This investigation presents a support vector regression (SVR)-based load forecasting model which hybridizes the chaotic mapping function and quantum particle swarm optimization algorithm with a support vector regression model, namely the SVRCQPSO (support vector regression with chaotic quantum particle swarm optimization) model, to achieve more accurate forecasting performance. Experimental results indicate that the proposed SVRCQPSO model achieves more accurate forecasting results than other alternatives.

Possible sources of forecast errors generated by the global/regional assimilation and prediction system for landfalling tropical cyclones. Part I: Initial uncertainties

This paper investigates the possible sources of errors associated with tropical cyclone (TC) tracks forecasted using the Global/Regional Assimilation and Prediction System (GRAPES). The GRAPES forecasts were made for 16 landfalling TCs in the western North Pacific basin during the 2008 and 2009 seasons, with a forecast length of 72 hours, and using the default initial conditions (“initials”, hereafter), which are from the NCEP-FNL dataset, as well as ECMWF initials. The forecasts are compared with ECMWF forecasts. The results show that in most TCs, the GRAPES forecasts are improved when using the ECMWF initials compared with the default initials. Compared with the ECMWF initials, the default initials produce lower intensity TCs and a lower intensity subtropical high, but a higher intensity South Asia high and monsoon trough, as well as a higher temperature but lower specific humidity at the TC center. Replacement of the geopotential height and wind fields with the ECMWF initials in and around the TC center at the initial time was found to be the most efficient way to improve the forecasts. In addition, TCs that showed the greatest improvement in forecast accuracy usually had the largest initial uncertainties in TC intensity and were usually in the intensifying phase. The results demonstrate the importance of the initial intensity for TC track forecasts made using GRAPES, and indicate the model is better in describing the intensifying phase than the decaying phase of TCs. Finally, the limit of the improvement indicates that the model error associated with GRAPES forecasts may be the main cause of poor forecasts of landfalling TCs. Thus, further examinations of the model errors are required.

Learning for improvement

Learning for improvement

Probabilistic Wind Generation Forecast Based on Sparse Bayesian Classification and Dempster–Shafer Theory

Probabilistic wind generation forecast results are crucial for power system operational dispatch. In this paper, a nonparametric approach for short-term probabilistic wind generation forecast based on the sparse Bayesian classification (SBC) and Dempster–Shafer theory (DST) is proposed. This approach is composed of the following four steps. 1) A spot forecast of wind generation is performed based on support vector machine (SVM). 2) The range of SVM forecast error is discretized into multiple intervals, and the conditional probability of each interval is estimated by a sparse Bayesian classifier. 3) DST is applied to combine the probabilities of all the intervals to form a unified probability distribution function (pdf) of the SVM forecast error. 4) The pdf of wind generation is achieved by combining the SVM wind generation spot forecast result and corresponding forecast error distribution. The distinguishing features of the proposed approach are as follows. 1) The approach is a nonparametric one and the forecast error caused by the misjudgement of probability distribution type can be avoided. 2) The proposed approach has good generalization capability by using the sparse learning mechanism. 3) The range constraint of wind generation can be systematically considered in the approach by applying DST. Tests on a 74-MW wind farm illustrate the improvement of spot forecast accuracy of the proposed approach and validate that this paper provides better calibrated and sharper probabilistic forecasts than the empirical approach and quantile regression model. Comparison to the state-of-the-art of the Global Energy Forecasting Competition 2014 demonstrates the probabilistic forecast performance of the proposed approach.

Can tourism confidence index improve tourism demand forecasts?

PurposeThe link between confidence and economic decisions has been widely covered in the economic literature, yet it is still an unexplored field in tourism. The purpose of this paper is to address this gap, and investigate benefits in forecast accuracy that can be achieved by combining the UNWTO Tourism Confidence Index (TCI) with statistical forecasts.Design/methodology/approachResearch is conducted in a real-life setting, using UNWTO unique data sets of tourism indicators. UNWTO TCI is pooled with statistical forecasts using three distinct approaches. Forecasts efficiency is assessed in terms of accuracy gains and capability to predict turning points in alternative scenarios, including one of the hardest crises the tourism sector ever experienced.FindingsResults suggest that the TCI provides meaningful indications about the sign of future growth in international tourist arrivals, and point to an improvement of forecast accuracy, when the index is used in combination with statistical forecasts. Still, accuracy gains vary greatly across regions and can hardly be generalised. Findings provide meaningful directions to tourism practitioners on the use opportunity cost to produce short-term forecasts using both approaches.Practical implicationsEmpirical evidence suggests that a confidence index should not be collected as input to improve their forecasts. It remains a valuable instrument to supplement official statistics, over which it has the advantage of being more frequently compiled and more rapidly accessible. It is also of particular importance to predict changes in the business climate and capture turning points in a timely fashion, which makes it an extremely valuable input for operational and strategic decisions.Originality/valueThe use of sentiment indexes as input to forecasting is an unexplored field in the tourism literature.

Observation impact analysis methods for storm surge forecasting systems

This paper presents a simple method for estimating the impact of assimilating individual or group of observations on forecast accuracy improvement. This method is derived from the nsemble-based observation impact analysis method of Liu and Kalnay (Q J R Meteorol Soc 134:1327–1335, 2008). The method described here is different in two ways from their method. Firstly, it uses a quadratic function of model-minus-observation residuals as a measure of forecast accuracy, instead of model-minus-analysis. Secondly, it simply makes use of time series of observations and the corresponding model output generated without data assimilation. These time series are usually available in an operational database. Hence, it is simple to implement. It can be used before any data assimilation is implemented. Therefore, it is useful as a design tool of a data assimilation system, namely for selecting which observations to assimilate. The method can also be used as a diagnostic tool, for example, to assess if all observation contributes positively to the accuracy improvement. The method is applicable for systems with stationary error process and fixed observing network. Using twin experiments with a simple one-dimensional advection model, the method is shown to work perfectly in an idealized situation. The method is used to evaluate the observation impact in the operational storm surge forecasting system based on the Dutch Continental Shelf Model version 5 (DCSMv5).

Short-Term GDP Forecasting With a Mixed-Frequency Dynamic Factor Model With Stochastic Volatility

In this article, we develop a mixed frequency dynamic factor model in which the disturbances of both the latent common factor and of the idiosyncratic components have time-varying stochastic volatilities. We use the model to investigate business cycle dynamics in the euro area and present three sets of empirical results. First, we evaluate the impact of macroeconomic releases on point and density forecast accuracy and on the width of forecast intervals. Second, we show how our setup allows to make a probabilistic assessment of the contribution of releases to forecast revisions. Third, we examine point and density out of sample forecast accuracy. We find that introducing stochastic volatility in the model contributes to an improvement in both point and density forecast accuracy. Supplementary materials for this article are available online.

Probing NWP model deficiencies by statistical postprocessing

This article describes a Model Output Statistics (MOS) framework for wind speed forecast accuracy improvement. The approach is used to identify non‐intuitive explanatory value from a diagnostic variable in an operational numerical weather prediction (NWP) model generating global weather forecasts four times daily, with numerous users worldwide. The analysis is based on two years of hourly wind speed time series measured at three locations: offshore, in coastal flat terrain, and inland in complex topography. Together with wind speed and wind direction predictors derived from the prognostic wind field, the lifted index NWP model diagnostic is found to perform well as statistical wind speed forecast model predictor.

Forecasting unemployment across countries: The ins and outs

This paper evaluates the flow approach to unemployment forecasting proposed by Barnichon and Nekarda (2012) for a set of OECD countries characterized by very different labor markets. We find that the flow approach yields substantial improvements in forecast accuracy over professional forecasts for all countries, with especially large improvements at longer horizons (one-year ahead forecasts) for European countries. Moreover, the flow approach has the highest predictive ability during recessions and turning points, when unemployment forecasts are most valuable.

Analysis and forecasting of demand during promotions for perishable items

This study examines promotions for perishable products in a retail environment. We analyze the impact of relative price discounts on product sales during a promotion and shed light on how to build models to forecast promotional demand for perishable products. Preliminary analyses, based on regression models and a large dataset from a retailer, do not reveal conclusive evidence for the presence of threshold and/or saturation levels for price discounts for perishable products. A potential explanation comes from the observation that, although products like desserts on average allow 1,5 weeks time-to-consume, their sales during promotions on average are equal to 14 weeks of regular sales. This suggests that the success of a promotion is not so much determined by the restriction to stockpile (due to the short time-to-consume) but by the emergence of substitution effects (consumers switching between different products of the same category). We develop and test different models to forecast the demand during a promotion, including a moving average forecast and several regression models. Within the class of regression models we find that modeling threshold and saturation effects leads to worse forecasting performance than modeling price reductions linearly or quadratically. The largest improvements in forecast accuracy are gained by distinguishing between routine and non-routine product categories. Routine categories with routine demand processes and a large number of observations perform best when applying a regression based on direct observations of the product category, whilst non-routine categories benefit from a regression which also uses observations from other product categories.

Forecasting Inflation: Phillips Curve Effects on Services Price Measures

We estimate an empirical model of inflation that exploits a Phillips curve relationship between a measure of unemployment and a subaggregate measure of inflation (services). We generate an aggregate inflation forecast from forecasts of the goods subcomponent separate from the services subcomponent, and compare the aggregated forecast to the leading time-series univariate and standard Phillips curve forecasting models. Our results indicate notable improvements in forecasting accuracy statistics for models that exploit relationships between services inflation and the unemployment rate. In addition, models of services inflation using the short-term unemployment rate (less than 27 weeks) as the real economic indicator display additional modest forecast accuracy improvements.

New product introduction and supplier integration in sales and operations planning

Purpose – The purpose of this paper is to investigate the implementation and performance benefits of sales and operations planning (S & OP) within organizations in Asia Pacific. Design/methodology/approach – A case study method was used, with two companies selected. The first company had recently commenced S & OP and applied it to facilitate new product introduction, while the second had integrated its supplier into an existing S & OP program. Supply chain performance data were collected and analyzed in the context of an S & OP maturity framework. Findings – Both cases show significant improvements in supply chain performance. In one case, the implementation of a common form of S & OP resulted in a 67 percent reduction in order lead time for newly introduced products. The second case demonstrated a 30 percent reduction in inventory levels and a 52 percent improvement in forecast accuracy through more advanced S & OP processes. Research limitations/implications – This paper studies just two companies and is not intended to be representative of outcomes at all companies implementing S & OP. Further studies are required for a more generalized picture of S & OP implementations in the Asia Pacific region to emerge. Practical implications – The findings illustrate the potential quantitative benefits of adopting S & OP and the circumstances under which these benefits may be achieved. The results are also supportive of the notion of a maturity model for S & OP implementations. Originality/value – This paper strengthens the link between practitioner and academic literature by providing empirical evidence of the benefits of S & OP. Furthermore, the findings are derived from the Asia Pacific region for which there have been few academic studies on S & OP to date.

Improving real-time inflow forecasting into hydropower reservoirs through a complementary modelling framework

Abstract. Accuracy of reservoir inflow forecasts is instrumental for maximizing the value of water resources and benefits gained through hydropower generation. Improving hourly reservoir inflow forecasts over a 24 h lead time is considered within the day-ahead (Elspot) market of the Nordic exchange market. A complementary modelling framework presents an approach for improving real-time forecasting without needing to modify the pre-existing forecasting model, but instead formulating an independent additive or complementary model that captures the structure the existing operational model may be missing. We present here the application of this principle for issuing improved hourly inflow forecasts into hydropower reservoirs over extended lead times, and the parameter estimation procedure reformulated to deal with bias, persistence and heteroscedasticity. The procedure presented comprises an error model added on top of an unalterable constant parameter conceptual model. This procedure is applied in the 207 km2 Krinsvatn catchment in central Norway. The structure of the error model is established based on attributes of the residual time series from the conceptual model. Besides improving forecast skills of operational models, the approach estimates the uncertainty in the complementary model structure and produces probabilistic inflow forecasts that entrain suitable information for reducing uncertainty in the decision-making processes in hydropower systems operation. Deterministic and probabilistic evaluations revealed an overall significant improvement in forecast accuracy for lead times up to 17 h. Evaluation of the percentage of observations bracketed in the forecasted 95 % confidence interval indicated that the degree of success in containing 95 % of the observations varies across seasons and hydrologic years.