Short-term electricity demand forecasting using double seasonal exponential smoothing

This paper considers univariate online electricity demand forecasting for lead times from a half-hour-ahead to a day-ahead. A time series of demand recorded at half-hourly intervals contains more than one seasonal pattern. A within-day seasonal cycle is apparent from the similarity of the demand profile from one day to the next, and a within-week seasonal cycle is evident when one compares the demand on the corresponding day of adjacent weeks. There is strong appeal in using a forecasting method that is able to capture both seasonalities. The multiplicative seasonal ARIMA model has been adapted for this purpose. In this paper, we adapt the Holt–Winters exponential smoothing formulation so that it can accommodate two seasonalities. We correct for residual autocorrelation using a simple autoregressive model. The forecasts produced by the new double seasonal Holt–Winters method outperform those from traditional Holt–Winters and from a well-specified multiplicative double seasonal ARIMA model.     

Triple seasonal methods for short-term electricity demand forecasting

Online short-term load forecasting is needed for the real-time scheduling of electricity generation. Univariate methods have been developed that model the intraweek and intraday seasonal cycles in intraday load data. Three such methods, shown to be competitive in recent empirical studies, are double seasonal ARMA, an adaptation of Holt–Winters exponential smoothing for double seasonality, and another, recently proposed, exponential smoothing method. In multiple years of load data, in addition to intraday and intraweek cycles, an intrayear seasonal cycle is also apparent. We extend the three double seasonal methods in order to accommodate the intrayear seasonal cycle. Using six years of British and French data, we show that for prediction up to a day-ahead the triple seasonal methods outperform the double seasonal methods, and also a univariate neural network approach. Further improvement in accuracy is produced by using a combination of the forecasts from two of the triple seasonal methods.     

Use of individual and group seasonal indices in subaggregate demand forecasting

An individual seasonal indices (ISI) method and two group seasonal indices (GSI) methods proposed in the literature are compared, based on two models. Rules have been established to choose between these methods and insights are gained on the conditions under which one method outperforms the others. Simulation findings confirm that using the rules improves forecasting accuracy against universal application of these methods.     

Forecasting seasonal time series data: a Bayesian model averaging approach

A flexible Bayesian periodic autoregressive model is used for the prediction of quarterly and monthly time series data. As the unknown autoregressive lag order, the occurrence of structural breaks and their respective break dates are common sources of uncertainty these are treated as random quantities within the Bayesian framework. Since no analytical expressions for the corresponding marginal posterior predictive distributions exist a Markov Chain Monte Carlo approach based on data augmentation is proposed. Its performance is demonstrated in Monte Carlo experiments. Instead of resorting to a model selection approach by choosing a particular candidate model for prediction, a forecasting approach based on Bayesian model averaging is used in order to account for model uncertainty and to improve forecasting accuracy. For model diagnosis a Bayesian sign test is introduced to compare the predictive accuracy of different forecasting models in terms of statistical significance. In an empirical application, using monthly unemployment rates of Germany, the performance of the model averaging prediction approach is compared to those of model selected Bayesian and classical (non)periodic time series models.     

Improving demand forecasting accuracy using nonlinear programming software

We address the problem of forecasting real time series with a proportion of zero values and a great variability among the nonzero values. In order to calculate forecasts for a time series, the model coefficients must be estimated. The appropriate choice of values for the smoothing parameters in exponential smoothing methods relies on the minimization of the fitting errors of historical data. We adapt the generalized Holt–Winters formulation so that it can consider the starting values of the local components of level, trend and seasonality as decision variables of the nonlinear programming problem associated with this forecasting procedure. A spreadsheet model is used to solve the problems of optimization efficiently. We show that our approach produces accurate forecasts with little data per product.     

Two-sample Kolmogorov-Smirnov test using a Bayesian nonparametric approach

In this paper, a Bayesian nonparametric approach to the two-sample problem is proposed. Given two samples \(\text{X} = {X_1}, \ldots ,{X_{m1}}\;\mathop {\text~}\limits^{i.i.d.} F\) and \(Y = {Y_1}, \ldots ,{Y_{{m_2}}}\mathop {\text~}\limits^{i.i.d.} G\), with F and G being unknown continuous cumulative distribution functions, we wish to test the null hypothesis H ₀: F = G. The method is based on computing the Kolmogorov distance between two posterior Dirichlet processes and comparing the results with a reference distance. The parameters of the Dirichlet processes are selected so that any discrepancy between the posterior distance and the reference distance is related to the difference between the two samples. Relevant theoretical properties of the procedure are also developed. Through simulated examples, the approach is compared to the frequentist Kolmogorov–Smirnov test and a Bayesian nonparametric test in which it demonstrates excellent performance.     

Electricity consumption prediction using a neural-network-based grey forecasting approach

Electricity consumption is an important economic index and plays a significant role in drawing up an energy development policy for each country. Multivariate techniques and time-series analysis have been proposed to deal with electricity consumption forecasting, but a large amount of historical data is required to obtain accurate predictions. The grey forecasting model attracted researchers by its ability to characterize an uncertain system effectively with a limited number of samples. GM(1,1) is the most frequently used grey forecasting model, but its developing coefficient and control variable were dependent on the background value that is not easy to be determined, whereas a neural-network-based GM(1,1) model called NNGM(1,1) has been presented to resolve this troublesome problem. This study has applied NNGM(1,1) to electricity consumption and has examined its forecasting ability on electricity consumption using sample data from the Turkish Ministry of Energy and Natural Resources and the Asia–Pacific Economic Cooperation energy database. Experimental results demonstrate that NNGM(1,1) performs well.     

Bayesian tail‐risk forecasting using realized GARCH

A realized generalized autoregressive conditional heteroskedastic (GARCH) model is developed within a Bayesian framework for the purpose of forecasting value at risk and conditional value at risk. Student‐t and skewed‐t return distributions are combined with Gaussian and student‐t distributions in the measurement equation to forecast tail risk in eight international equity index markets over a 4‐year period. Three realized measures are considered within this framework. A Bayesian estimator is developed that compares favourably, in simulations, with maximum likelihood, both in estimation and forecasting. The realized GARCH models show a marked improvement compared with ordinary GARCH for both value‐at‐risk and conditional value‐at‐risk forecasting. This improvement is consistent across a variety of data and choice of distributions. Realized GARCH models incorporating a skewed student‐t distribution for returns are favoured overall, with the choice of measurement equation error distribution and realized measure being of lesser importance. Copyright © 2017 John Wiley & Sons, Ltd.     

A two-step method for forecasting spare parts demand using information on component repairs

Forecasting spare parts demand is notoriously difficult, as demand is typically intermittent and lumpy. Specialized methods such as that by Croston are available, but these are not based on the repair operations that cause the intermittency and lumpiness of demand. In this paper, we do propose a method that, in addition to the demand for spare parts, considers the type of component repaired. This two-step forecasting method separately updates the average number of parts needed per repair and the number of repairs for each type of component. The method is tested in an empirical, comparative study for a service provider in the aviation industry. Our results show that the two-step method is one of the most accurate methods, and that it performs considerably better than Croston’s method. Moreover, contrary to other methods, the two-step method can use information on planned maintenance and repair operations to reduce forecasts errors by up to 20%. We derive further analytical and simulation results that help explain the empirical findings.     

An improved seasonal rolling grey forecasting model using a cycle truncation accumulated generating operation for traffic flow

Accurate real-time prediction of urban traffic flows is one of the most important problems in traffic management and control optimization research. Short-term traffic flow has complex stochastic and nonlinear characteristics, and it shows a similar seasonality within intraday and weekly trends. Based on these properties, we propose an improved binding cycle truncation accumulated generating operation seasonal grey rolling forecasting model. In the new model, the traffic flow sequence of seasonal fluctuation is converted to a flat sequence using the cycle truncation accumulated generating operation. Then, grey modeling of the cycle truncation accumulated generating operation sequence weakens the stochastic disturbances and highlights the intrinsic grey exponential law after the sequence is accumulated. Finally, rolling forecasts of the limited data reflect the new information priority and timeliness of the grey prediction. Two numerical traffic flow examples from China and Canada, including four groups at different time intervals (1 h, 15 min, 10 min, and 5 min), are used to verify the performance of the new model under different traffic flow conditions. The prediction results show that the model has good adaptability and stability and can effectively predict the seasonal variations in traffic flow. In 15 or 10 min traffic flow forecasts, the proposed model shows better performance than the autoregressive moving average model, wavelet neural network model and seasonal discrete grey forecasting model.     

A new approach of forecasting intermittent demand for spare parts inventories in the process industries

Accurate demand forecasting is of vital importance in inventory management of spare parts in process industries, while the intermittent nature makes demand forecasting for spare parts especially difficult. With the wide application of information technology in enterprise management, more information and data are now available to improve forecasting accuracy. In this paper, we develop a new approach for forecasting the intermittent demand of spare parts. The described approach provides a mechanism to integrate the demand autocorrelated process and the relationship between explanatory variables and the nonzero demand of spare parts during forecasting occurrences of nonzero demands over lead times. Two types of performance measures for assessing forecast methods are also described. Using data sets of 40 kinds of spare parts from a petrochemical enterprise in China, we show that our method produces more accurate forecasts of lead time demands than do exponential smoothing, Croston's method and Markov bootstrapping method.     

Experimental study of a Bayesian method for daily electricity load forecasting

The method of Bayesian model discrimination is investigated for the possible contributions it may provide in the area of automatically forecasting the daily electricity demand cycle. A set of differing demand models have probabilities attached to them in such a way that these would be continuously updated with the available data and the actual forecasts obtained as expectations across all the models. Simulation experiments indicate significantly improved forecasting performance over a commonly used rescaling type of approach. Some practical issues in implementation are discussed.     

Modeling special-day effects for forecasting intraday electricity demand

We propose and apply a novel approach for modeling special-day effects to predict electricity demand in Korea. Notably, we model special-day effects on an hourly rather than a daily basis. Hourly specified predictor variables are implemented in the regression model with a seasonal autoregressive moving average (SARMA) type error structure in order to efficiently reflect the special-day effects. The interaction terms between the hour-of-day effects and the hourly based special-day effects are also included to capture the unique intraday patterns of special days more accurately. The multiplicative SARMA mechanism is employed in order to identify the double seasonal cycles, namely, the intraday effect and the intraweek effect. The forecast results of the suggested model are evaluated by comparing them with those of various benchmark models for the following year. The empirical results indicate that the suggested model outperforms the benchmark models for both special- and non-special day predictions.     

Supply chain network downsizing with product line pruning using a new demand substitution

This paper presents an optimization model for downsizing a multi-product supply chain facing bankruptcy risk, where multi-functional production facilities are shared for producing a group of substitutable products. In order to determine the potential demand after discontinuation of certain product types, the proposed mixed integer programming model incorporates a new general formulation of demand substitution, which allows arbitrary demand diversion and arbitrary replacement rates between products under investigation. The new substitution formulation enables considering uneven substitutions for supply chain network design and therefore it extends currently used substitution approaches. A series of systematically generated downsizing cases confirm the validity of the here proposed approach, and illustrate how downsized chains can successfully adopt to the new circumstances.     

Real options approach-based demand forecasting method for a range of products with highly volatile and correlated demand

To achieve a competitive edge needed for marketing highly competitive products, modern enterprises have actively sought to provide the marketplace with an expansive range of products with high random volatility of demand and correlations between demands of product. Consequently, traditional forecasting methods for separately forecasting demand for these products are likely to yield significant deviations. Therefore, this study develops a real options approach-based forecasting model to accurately predict future demand for a given range of products with highly volatile and correlated demand. Additionally, this study also proposes using Monte Carlo simulation to solve the demand forecasting model. The real options approach associated with Monte Carlo simulation not only deals effectively with random variation involving a particular demand stochastic diffusion process, but can handle the correlations in product demand.     

Neural network forecasting for seasonal and trend time series

Neural networks have been widely used as a promising method for time series forecasting. However, limited empirical studies on seasonal time series forecasting with neural networks yield mixed results. While some find that neural networks are able to model seasonality directly and prior deseasonalization is not necessary, others conclude just the opposite. In this paper, we investigate the issue of how to effectively model time series with both seasonal and trend patterns. In particular, we study the effectiveness of data preprocessing, including deseasonalization and detrending, on neural network modeling and forecasting performance. Both simulation and real data are examined and results are compared to those obtained from the Box–Jenkins seasonal autoregressive integrated moving average models. We find that neural networks are not able to capture seasonal or trend variations effectively with the unpreprocessed raw data and either detrending or deseasonalization can dramatically reduce forecasting errors. Moreover, a combined detrending and deseasonalization is found to be the most effective data preprocessing approach.     

A mega-trend-diffusion grey forecasting model for short-term manufacturing demand

Accurate short-term demand forecasting is critical for developing effective production plans; however, a short forecasting period indicates that the product demands are unstable, rendering tracking of product development trends difficult. Determining the actual developing data patterns by using forecasting models generated using historical observations is difficult, and the forecasting performance of such models is unfavourable, whereas using the latest limited data for forecasting can improve management efficiency and maintain the competitive advantages of an enterprise. To solve forecasting problems related to a small data set, this study applied an adaptive grey model for forecasting short-term manufacturing demand. Experiments involving the monthly demand data for thin film transistor liquid crystal display panels and wafer-level chip-scale packaging process data showed that the proposed grey model produced favourable forecasting results, indicating its appropriateness as a short-term forecasting tool for small data sets.     

‘Horses for Courses’ in demand forecasting

Forecasting as a scientific discipline has progressed a lot in the last 40 years, with Nobel prizes being awarded for seminal work in the field, most notably to Engle, Granger and Kahneman. Despite these advances, even today we are unable to answer a very simple question, the one that is always the first tabled during discussions with practitioners: “what is the best method for my data?”. In essence, as there are horses for courses, there must also be forecasting methods that are more tailored to some types of data, and, therefore, enable practitioners to make informed method selection when facing new data. The current study attempts to shed light on this direction via identifying the main determinants of forecasting accuracy, through simulations and empirical investigations involving 14 popular forecasting methods (and combinations of them), seven time series features (seasonality, trend, cycle, randomness, number of observations, inter-demand interval and coefficient of variation) and one strategic decision (the forecasting horizon). Our main findings dictate that forecasting accuracy is influenced as follows: (a) for fast-moving data, cycle and randomness have the biggest (negative) effect and the longer the forecasting horizon, the more accuracy decreases; (b) for intermittent data, inter-demand interval has bigger (negative) impact than the coefficient of variation; and (c) for all types of data, increasing the length of a series has a small positive effect.     

Selecting the best periodic inventory control and demand forecasting methods for low demand items

The (s,S) form of the periodic review inventory control system has been claimed theoretically to be the best for the management of items of low and intermittent demand. Various heuristic procedures have been put forward, usually justified on the basis of generated data with known properties. Some stock controllers also have other simple rules which they employ and which are rarely seen in the literature. Determining how to forecast future demands is also a major problem in the area. The research described in this paper compares various periodic inventory policies as well as some forecasting methods and attempts to determine which are best for low and intermittent demand items. It evaluates the alternative methods on some long series of daily demands for low demand items for a typical spare parts depot.     

Application of machine learning techniques for supply chain demand forecasting

Full collaboration in supply chains is an ideal that the participant firms should try to achieve. However, a number of factors hamper real progress in this direction. Therefore, there is a need for forecasting demand by the participants in the absence of full information about other participants’ demand. In this paper we investigate the applicability of advanced machine learning techniques, including neural networks, recurrent neural networks, and support vector machines, to forecasting distorted demand at the end of a supply chain (bullwhip effect). We compare these methods with other, more traditional ones, including naïve forecasting, trend, moving average, and linear regression. We use two data sets for our experiments: one obtained from the simulated supply chain, and another one from actual Canadian Foundries orders. Our findings suggest that while recurrent neural networks and support vector machines show the best performance, their forecasting accuracy was not statistically significantly better than that of the regression model.