Demand Forecasting Intermittent and Lumpy Time Series: Comparing Statistical, Machine Learning and Deep Learning Methods

Forecasting intermittent and lumpy demand is challenging. Demand occurs only sporadically and, when it does, it can vary considerably. Forecast errors are costly, resulting in obsolescent stock or unmet demand. Methods from statistics, machine learning and deep learning have been used to predict such demand patterns. Traditional accuracy metrics are often employed to evaluate the forecasts, however these come with major drawbacks such as not taking horizontal and vertical shifts over the forecasting horizon into account, or indeed stock-keeping or opportunity costs. This results in a disadvantageous selection of methods in the context of intermittent and lumpy demand forecasts. In our study, we compare methods from statistics, machine learning and deep learning by applying a novel metric called Stock-keeping-oriented Prediction Error Costs (SPEC), which overcomes the drawbacks associated with traditional metrics. Taking the SPEC metric into account, the Croston algorithm achieves the best result, just ahead of a Long Short-Term Memory Neural Network.

[1]  C. Holt.  Author's retrospective on ‘Forecasting seasonals and trends by exponentially weighted moving averages’ , 2004 .

[2]  Aris A. Syntetos,et al.  On the categorization of demand patterns , 2005, J. Oper. Res. Soc..

[3]  Nikolaos Kourentzes,et al.  Intermittent demand forecasts with neural networks , 2013 .

[4]  Anders Segerstedt,et al.  Evaluation of forecasting error measurements and techniques for intermittent demand , 2010 .

[5]  A. B. Oktay,et al.  Air Pollution Forecasting with Random Forest Time Series Analysis , 2018, 2018 International Conference on Artificial Intelligence and Data Processing (IDAP).

[6]  Yuliang Shi,et al.  Electricity Consumption Prediction Using XGBoost Based on Discrete Wavelet Transform , 2017 .

[7]  T. Williams.  Stock Control with Sporadic and Slow-Moving Demand , 1984 .

[8]  Peter R. Winters,et al.  Forecasting Sales by Exponentially Weighted Moving Averages , 1960 .

[9]  D. Ivanov,et al.  Global Supply Chain and Operations Management , 2021, Springer Texts in Business and Economics.

[10]  Ping-Feng Pai,et al.  Time series forecasting by a seasonal support vector regression model , 2010, Expert Syst. Appl..

[11]  Sebastián Maldonado,et al.  Hierarchical time series forecasting via Support Vector Regression in the European Travel Retail Industry , 2019, Expert Syst. Appl..

[12]  J. Boylan,et al.  On the bias of intermittent demand estimates , 2001 .

[13]  Heeyoung Kim,et al.  A new metric of absolute percentage error for intermittent demand forecasts , 2016 .

[14]  J. Boylan,et al.  The accuracy of intermittent demand estimates , 2005 .

[15]  J. D. Croston.  Forecasting and Stock Control for Intermittent Demands , 1972 .

[16]  Zhongsheng Hua,et al.  A hybrid support vector machines and logistic regression approach for forecasting intermittent demand of spare parts , 2006, Appl. Math. Comput..

[17]  Tanveer Ahmad,et al.  Nonlinear autoregressive and random forest approaches to forecasting electricity load for utility energy management systems , 2019, Sustainable Cities and Society.

[18]  Slawek Smyl,et al.  A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting , 2020, International Journal of Forecasting.

[19]  Geoffrey E. Hinton,et al.  A Learning Algorithm for Boltzmann Machines , 1985, Cogn. Sci..

[20]  Andrew Lim,et al.  A greedy aggregation-decomposition method for intermittent demand forecasting in fashion retailing , 2018, Eur. J. Oper. Res..

[21]  Richard T. Watson,et al.  Analyzing the Past to Prepare for the Future: Writing a Literature Review , 2002, MIS Q..

[22]  Rob J. Hyndman,et al.  A note on the validity of cross-validation for evaluating autoregressive time series prediction , 2018, Comput. Stat. Data Anal..

[23]  K. Nikolopoulos,et al.  The theta model: a decomposition approach to forecasting , 2000 .

[24]  T. Willemain,et al.  A new approach to forecasting intermittent demand for service parts inventories , 2004 .

[25]  Mohsen Yousefi,et al.  An optimized model using LSTM network for demand forecasting , 2020, Comput. Ind. Eng..

[26]  Konstantinos Nikolopoulos,et al.  We need to talk about intermittent demand forecasting , 2020, Eur. J. Oper. Res..

[27]  Aris A. Syntetos,et al.  Forecasting and Inventory Performance in a Two-Stage Supply Chain with ARIMA(0,1,1) Demand: Theory and Empirical Analysis , 2013 .

[28]  Na Wang,et al.  Review of Croston's method for intermittent demand forecasting , 2012, 2012 9th International Conference on Fuzzy Systems and Knowledge Discovery.

[29]  Yair Levy,et al.  A Systems Approach to Conduct an Effective Literature Review in Support of Information Systems Research , 2006, Informing Sci. Int. J. an Emerg. Transdiscipl..

[30]  Ravi Sankar,et al.  Time Series Prediction Using Support Vector Machines: A Survey , 2009, IEEE Computational Intelligence Magazine.

[31]  Xue Ben,et al.  Deep Transformer Models for Time Series Forecasting: The Influenza Prevalence Case , 2020, ArXiv.

[32]  Rob J Hyndman,et al.  Another look at measures of forecast accuracy , 2006 .

[33]  Niklas Kühl,et al.  A New Metric for Lumpy and Intermittent Demand Forecasts: Stock-keeping-oriented Prediction Error Costs , 2020, HICSS.

[34]  T. Willemain,et al.  Forecasting intermittent demand in manufacturing: a comparative evaluation of Croston's method , 1994 .

[35]  Evangelos Spiliotis,et al.  The M4 Competition: 100,000 time series and 61 forecasting methods , 2020 .

[36]  Rob J. Hyndman,et al.  A note on the categorization of demand patterns , 2006, J. Oper. Res. Soc..

[37]  Anders Segerstedt,et al.  Inventory control with a modified Croston procedure and Erlang distribution , 2004 .

[38]  Yan Wang,et al.  Forecasting method of stock market volatility in time series data based on mixed model of ARIMA and XGBoost , 2020, China Communications.

[39]  Rita Gamberini,et al.  Forecasting of Sporadic Demand Patterns with Seasonality and Trend Components: An Empirical Comparison between Holt-Winters and (S)ARIMA Methods , 2010 .

[40]  Chen-Fu Chien,et al.  A Hybrid Forecasting Framework with Neural Network and Time-Series Method for Intermittent Demand in Semiconductor Supply Chain , 2018, APMS.

[41]  Sven F. Crone,et al.  Artificial Neural Networks for Time Series Prediction - A Novel Approach to Inventory Management Using Asymmetric Cost Functions , 2003, IC-AI.

[42]  Adriano O. Solis,et al.  The Accuracy of Non‐traditional versus Traditional Methods of Forecasting Lumpy Demand , 2012 .

[43]  R. Quintana,et al.  Lumpy demand forecasting using neural networks , 2001, PICMET '01. Portland International Conference on Management of Engineering and Technology. Proceedings Vol.1: Book of Summaries (IEEE Cat. No.01CH37199).