A modified Teunter-Syntetos-Babai method for intermittent demand forecasting

Intermittent demand refers to the specific demand pattern with frequent periods of zero demand. It occurs in a variety of industries including industrial equipment, automotive and specialty chemicals. In some industries or some sectors of industry, even majority of products are in intermittent demand pattern. Due to the usually small and highly variable demand sizes, accurate forecasting of intermittent demand has always been challenging.However, accurate forecasting of intermittent demand is critical to the effective inventory management. In this study we present a band new method - modified TSB method for the forecasting of intermittent demand. The proposed method is based on TSB method, and adopts similar strategy, which has been used in mSBA method to update demand interval and demand occurrence probability when current demand is zero. To evaluate the proposed method, 16289 daily demand records from the M5 data set that are identified as intermittent demands according to two criteria, and an empirical data set consisting three years’ monthly demand history of 1718 medicine products are used. The proposed mTSB method achieves the best results on MASE and RMASE among all comparison methods on the M5 data set. On the empirical data set, the study shows that mTSB attains an ME of 0.07, which is the best among six comparison methods. Additionally, on the MSE measurement, mTSB shows a similar result as SES, both of which outperform other methods.     

Developing grey prediction with Fourier series using genetic algorithms for tourism demand forecasting

Quality & Quantity - Predicting the number of foreign tourists is significant for governments in devising development policies for tourism demand. Time series related to tourism often feature...     

Exponentially smoothed regression analysis for demand forecasting

This article proposes a new technique for estimating trend and multiplicative seasonality in time series data. The technique is computationally quite straightforward and gives better forecasts (in a sense described below) than other commonly used methods. Like many other methods, the one presented here is basically a decomposition technique, that is, it attempts to isolate and estimate the several subcomponents in the time series. It draws primarily on regression analysis for its power and has some of the computational advantages of exponential smoothing. In particular, old estimates of base, trend, and seasonality may be smoothed with new data as they occur. The basic technique was developed originally as a way to generate initial parameter values for a Winters exponential smoothing model [4], but it proved to be a useful forecasting method in itself.The objective in all decomposition methods is to separate somehow the effects of trend and seasonality in the data, so that the two may be estimated independently. When seasonality is modeled with an additive form (Datum = Base + Trend + Seasonal Factor), techniques such as regression analysis with dummy variables or ratio-to-moving-average techniques accomplish this task well. It is more common, however, to model seasonality as a multiplicative form (as in the Winters model, for example, where Datum = [Base + Trend] * Seasonal Factor). In this case, it can be shown that neither of the techniques above achieves a proper separation of the trend and seasonal effects, and in some instances may give highly misleading results. The technique described in this article attempts to deal properly with multiplicative seasonality, while remaining computationally tractable.The technique is built on a set of simple regression models, one for each period in the seasonal cycle. These models are used to estimate individual seasonal effects and then pooled to estimate the base and trend. As new data occur, they are smoothed into the least-squares formulas with computations that are quite similar to those used in ordinary exponential smoothing. Thus, the full least-squares computations are done only once, when the forecasting process is first initiated. Although the technique is demonstrated here under the assumption that trend is linear, the trend may, in fact, assume any form for which the curve-fitting tools are available (exponential, polynomial, etc.).The method has proved to be easy to program and execute, and computational experience has been quite favorable. It is faster than the RTMA method or regression with dummy variables (which requires a multiple regression routine), and it is competitive with, although a bit slower than, ordinary triple exponential smoothing.     

Classification-based model selection in retail demand forecasting

Retailers supply a wide range of stock keeping units (SKUs), which may differ for example in terms of demand quantity, demand frequency, demand regularity, and demand variation. Given this diversity in demand patterns, it is unlikely that any single model for demand forecasting can yield the highest forecasting accuracy across all SKUs. To save costs through improved forecasting, there is thus a need to match any given demand pattern to its most appropriate prediction model. To this end, we propose an automated model selection framework for retail demand forecasting. Specifically, we consider model selection as a classification problem, where classes correspond to the different models available for forecasting. We first build labeled training data based on the models’ performances in previous demand periods with similar demand characteristics. For future data, we then automatically select the most promising model via classification based on the labeled training data. The performance is measured by economic profitability, taking into account asymmetric shortage and inventory costs. In an exploratory case study using data from an e-grocery retailer, we compare our approach to established benchmarks. We find promising results, but also that no single approach clearly outperforms its competitors, underlying the need for case-specific solutions.     

Using weather ensemble predictions in electricity demand forecasting

Weather forecasts are an important input to many electricity demand forecasting models. This study investigates the use of weather ensemble predictions in electricity demand forecasting for lead times from 1 to 10 days ahead. A weather ensemble prediction consists of 51 scenarios for a weather variable. We use these scenarios to produce 51 scenarios for the weather-related component of electricity demand. The results show that the average of the demand scenarios is a more accurate demand forecast than that produced using traditional weather forecasts. We use the distribution of the demand scenarios to estimate the demand forecast uncertainty. This compares favourably with estimates produced using univariate volatility forecasting methods.     

Restarting a forecasting system when demand suddenly changes

Exponential smoothing is commonly used in automatic forecasting systems. However, when only a small amount of historical data is relevant to future demands, the ad hoc startup methods used in exponential smoothing produce unexpected results. With large data sets, an exponentially smoothed average implicitly weights the data in a declining manner, similar to discounting. This pattern is important in that it minimizes a measure of forecast error. However, restarting with limited data distorts the weighting pattern. A new technique, termed the declining alpha method, is presented and shown to preserve the exponential weight pattern. The key is a formula that changes the smoothing constant each period. Examples are given to illustrate the method and contrast it to other startup techniques.     

Estimation of food demand in Turkey: method of an almost ideal demand system

In this study, data of the household income and consumption expenditure surveys conducted by the Turkish Statistical Institute for 1994 and 2003 years were used; income, price, and cross price elasticities under six aggregated product groups were estimated within the framework of the an almost ideal demand system approach for food expenditures; and estimation of household consumers’ food demand in Turkey was analyzed. According to the findings obtained, it was established that a price-bound change would appear in the food demand, and elasticities were calculated. Expenditures by product groups and price elasticities were obtained, and the product groups were aggregated as bread and cereals; meat, fish, and poultry; milk and dairy products, oil and egg; vegetables and fruits; various fast food and alcoholic and non-alcoholic beverages.     

Big data from dynamic pricing: A smart approach to tourism demand forecasting

Suppliers of tourist services continuously generate big data on ask prices. We suggest using this information, in the form of a price index, to forecast the occupation rates for virtually any time-space frame, provided that there are a sufficient number of decision makers “sharing” their pricing strategies on the web. Our approach guarantees great transparency and replicability, as big data from OTAs do not depend on search interfaces and can facilitate intelligent interactions between the territory and its inhabitants, thus providing a starting point for a smart decision-making process. We show that it is possible to obtain a noticeable increase in the forecasting performance by including the proposed leading indicator (price index) into the set of explanatory variables, even with very simple model specifications. Our findings offer a new research direction in the field of tourism demand forecasting leveraging on big data from the supply side.     

A note on forecasting international tourism demand in Spain

This paper considers the extent to which price and income proxy variables help in forecasting tourist demand in Spain. Contrary to some recent studies, we found that the inputs' contribution in terms of fitting and forecasting is nil when compared with alternative univariate models. Whether these findings are the results of the restrictions embedded in building the proxy inputs or in a poor specification of the dynamics of these models remains to be seen. We also contend that when dealing with medium, long-term forecasting comparisons, the use of the traditional aggregate accuracy measures like RMSE and MAPE help very little in discriminating among competing models. In these situations, predicted annual growth rates may be a better alternative.     

Long-term forecasting of fuel demand at theater entry points

Managing the distribution of fuel in theater requires Army fuel planners to forecast demand at the strategic level to ensure that fuel will be in the right place, at the right time, and in the amounts needed. This work presents a simulation approach to forecasting that accounts for the structure of the supply chain network when aggregating the demand of war fighters across the theater over the forecasting horizon. The resulting empirical distribution of demand at the theater entry point enables planners to identify forecast characteristics that impact their planning process, including the amplitudes and temporal positions of peaks in demand, and the estimated lead time to the point of use. Experimentation indicates that the forecasts are sensitive to the pattern of war fighter demand, the precise structure of the in-theater supply chain network, and the constraints and uncertainty present in the network, all of which are critical planning considerations.     

The use of advance order data in demand forecasting

This paper uses advance order data and historical demand data from a manufacturing shop and from a service operation to develop and test a forecasting methodology for predicting customer demand over a forecast horizon. The proposed methodology uses simple linear regression to model the relationship between a total demand ratio and a partial demand ratio. Comparison of the proposed model to a standard regression approach and a commonly used multiplicative model showed that the proposed model exhibited the greatest forecast accuracy.     

Combination of long term and short term forecasts, with application to tourism demand forecasting

Forecast combination is a well-established and well-tested approach for improving the forecasting accuracy. One beneficial strategy is to use constituent forecasts that have diverse information. In this paper we consider the idea of diversity being accomplished by using different time aggregations. For example, we could create a yearly time series from a monthly time series and produce forecasts for both, then combine the forecasts. These forecasts would each be tracking the dynamics of different time scales, and would therefore add diverse types of information. A comparison of several forecast combination methods, performed in the context of this setup, shows that this is indeed a beneficial strategy and generally provides a forecasting performance that is better than the performances of the individual forecasts that are combined.As a case study, we consider the problem of forecasting monthly tourism numbers for inbound tourism to Egypt. Specifically, we consider 33 individual source countries, as well as the aggregate. The novel combination strategy also produces a generally improved forecasting accuracy.     

Short interval forecasting of emergency phone call (911) work loads

There has been an increasing emphasis over the last 5 to 10 years to improve productivity in the Service Sector of the U.S. economy. Much of the improvement obtained by these managers has come about through better scheduling of the work force in these organizations. Effective scheduling of this personnel requires good estimates of demand, which may exhibit substantial variations between days for certain times of the year. The Indianapolis Police Department (IPD) Communications area is one such organization that exhibits varying workloads and is interested in improving staff scheduling of dispatch operators.This article explores the use of six different forecasting techniques for predicting daily emergency call workloads for the IPD's communications area. Historical call volume data are used to estimate the model parameters. A hold-out sample of five months compares forecasts and actual daily call levels. The forecast system utilizes a rolling horizon approach, where daily forecasts are made for the coming month from the end of the prior month. The forecast origin is then advanced to the end of the month, where the current month's actual call data are added to the historical database, new parameters are estimated, and then the next month's daily estimates are generated. Error measures of residual standard deviation, mean absolute percent error, and bias are used to measure performance. Statistical analyses are conducted to evaluate if significant differences in performance are present among the six models.The research presented in this article indicates that there are clearly significant differences in performance for the six models analyzed. These models were tailored to the specific structure and this work suggests that the short interval forecasting problems faced by many service organizations has several structural differences compared to the typical manufacturing firm in a made-to-stock environment. The results also suggests two other points. First, simple modeling approaches can perform well in complex environments that are present in many service organizations. Second, special tailoring of the forecasting model is necessary for many service firms. Historical data patterns for these organizations tend to be more complex than just trend and seasonal elements, which are normally tracked in smoothing models. These are important conclusions for both managers of operating systems and staff analysts supporting these operating systems. The design of an appropriate forecasting system to support effective staff planning must consider the nature, scope, and complexity of these environments.     

Testing a method to estimate county-level demand for educational attainment

Recent changes in federal support for local workforce training programs have created an increased need for methods of assessing demand for educational attainment in sub-state regions. This research extends current non-survey methods for estimating demand for educational attainment at the county level. It then evaluates the non-survey method on theoretical grounds, and compares survey and non-survey estimates for two West Virginia counties. We found significant differences between survey and non-survey results and conclude that current non-survey methodology will not consistently provide reasonably accurate estimates. We thus propose adjustments to this methodology that should improve its accuracy. Many, if not most, jurisdictions do not have the expertise or funds to carry out periodic employment surveys. Many local policymakers would therefore benefit from development of an inexpensive, straightforward non-survey methodology that provides reasonably accurate information on local employer demand for labor force characteristics such as educational attainment.     

An unobserved component model for multi-rate forecasting of telephone call demand: the design of a forecasting support system

An Unobserved Components (UC) Model based on an enhanced version of the Dynamic Harmonic Regression model, including new multi-rate and modulated cycle procedures, is used to develop a customised package for forecasting and signal extraction applied to hourly telephone call numbers made to Barclaycard plc. service centres, with a forecasting horizon of up to several weeks in advance. The paper outlines both the methodological and algorithmic aspects of the modelling, forecasting and signal extraction procedures, including the design and implementation of forecasting support software with a specially designed Graphical User Interface within the ^® computing environment. The forecasting performance is evaluated comprehensively in comparison with the well-known seasonal ARIMA approach.     

Money demand and the role of monetary indicators in forecasting euro area inflation

This paper examines the stability of money demand and the forecasting performances of a broad monetary aggregate (M3), excess liquidity and excess inflation in predicting euro area inflation. The out-of sample forecasting performances are compared to a widely used alternative, the spread of interest rates. The results indicate that the evolution of M3 is still in line with money demand, even when observations from the economic and financial crisis are included. Both excess measures and the spread are useful for predicting inflation.     

Improving demand forecasting in the air cargo handling industry: a case study

Air transportation plays a crucial role in the agile and dynamic environment of contemporary supply chains. This industry is characterised by high air cargo demand uncertainty, making forecasting extremely challenging. An in-depth case study has been undertaken in order to explore and untangle the factors influencing demand forecasting and consequently to improve the operational performance of an air cargo handling company. It has been identified that in practice, the demand forecasting process does not provide the necessary level of accuracy, to effectively cope with the high demand uncertainty. This has a negative impact on a whole range of air cargo operations, but especially on the management of the workforce, which is the most expensive resource in the air cargo handling industry. Besides forecast inaccuracy, a range of additional hidden factors that affect operations management have been identified. A number of recommendations have been made to improve demand forecasting and workforce management.     

Daily retail demand forecasting using machine learning with emphasis on calendric special days

Demand forecasting is an important task for retailers as it is required for various operational decisions. One key challenge is to forecast demand on special days that are subject to vastly different demand patterns than on regular days. We present the case of a bakery chain with an emphasis on special calendar days, for which we address the problem of forecasting the daily demand for different product categories at the store level. Such forecasts are an input for production and ordering decisions. We treat the forecasting problem as a supervised machine learning task and provide an evaluation of different methods, including artificial neural networks and gradient-boosted decision trees. In particular, we outline and discuss the possibility of formulating a classification instead of a regression problem. An empirical comparison with established approaches reveals the superiority of machine learning methods, while classification-based approaches outperform regression-based approaches. We also found that machine learning methods not only provide more accurate forecasts but are also more suitable for applications in a large-scale demand forecasting scenario that often occurs in the retail industry.     

Developing a decision rule to predict failure: The case of savings and loan associations

The relative costs of misclassifying institutions by their financial health is an issue that concerns researchers. In this paper, a model and decision rule are developed that improve the probability of identifying those Savings and Loans that are predicted not to fail, but are actually failing. For obvious reasons, stakeholders in those institutions are very much interested in avoiding this type I error. The study also makes available evidence that the examination of Z-scores can be useful in identifying other financial institutions that may experience financial failure.