ABSTRACT
A real world supply chain planning starts with the demand forecasting as a key input. In most scenarios, especially in fields like e-commerce where demand patterns are complex and are large scale, demand forecasting is done independent of supply chain constraints. There have been a plethora of methods, old and recent, for generating accurate forecasts. However, to the best of our knowledge, none of the methods take supply chain constraints into account during forecasting. In this paper, we are primarily interested in supply chain aware forecasting methods that does not impose any restrictions on demand forecasting process. We assume that the base forecasts follow a distribution from exponential family and are provided as input to supply chain planning by specifying the distribution form and parameters. With this in mind, following are the contributions of our paper. First, we formulate the supply chain aware forecast improvement of a base forecast as finding the game theoretically optimal parameters satisfying the supply chain constraints. Second, for regular distributions from exponential family, we show that this translates to projecting base forecast onto the (convex) set defined by supply constraints, which is at least as accurate as the base forecasts. Third, we note that using off the shelf convex solvers does not scale for large instances of supply chain, which is typical in e-commerce settings. We propose algorithms that scale better with problem size. We propose a general gradient descent based approach that works across different distributions from exponential family. We also propose a network flow based exact algorithm for Laplace distribution (which relates to mean absolute error, which is the most commonly used metric in forecasting). Finally, we substantiate the theoretical results with extensive experiments on a real life e-commerce data set as well as a range of synthetic data sets.
- Martín Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, et al. 2016. Tensorflow: Large-scale machine learning on heterogeneous distributed systems. arXiv preprint arXiv:1603.04467 (2016).Google Scholar
- Cruz E Borges, Yoseba K Penya, and Ivan Fernandez. 2012. Evaluating combined load forecasting in large power systems and smart grids. IEEE Transactions on Industrial Informatics, Vol. 9, 3 (2012), 1570--1577.Google ScholarCross Ref
- Stephen Boyd and Lieven Vandenberghe. 2004. Convex Optimization. Cambridge University Press, USA.Google ScholarDigital Library
- Stephen P. Boyd and Lieven Vandenberghe. 2014. Convex Optimization. Cambridge University Press. https://doi.org/10.1017/CBO9780511804441Google Scholar
- George B Dantzig and Philip Wolfe. 1960. Decomposition principle for linear programs. Operations research, Vol. 8, 1 (1960), 101--111.Google Scholar
- George B. Dantzig and Philip Wolfe. 1961. The Decomposition Algorithm for Linear Programs. Econometrica, Vol. 29, 4 (1961), 767--778. http://www.jstor.org/stable/1911818Google ScholarCross Ref
- Inderjit S. Dhillon and Joel A. Tropp. 2007. Matrix Nearness Problems with Bregman Divergences. SIAM J. Matrix Anal. Appl., Vol. 29, 4 (2007), 1120--1146. https://doi.org/10.1137/060649021Google ScholarDigital Library
- Yefim Dinitz. 2006. Dinitz' Algorithm: The Original Version and Even's Version. In Theoretical Computer Science, Essays in Memory of Shimon Even (Lecture Notes in Computer Science, Vol. 3895),, Oded Goldreich, Arnold L. Rosenberg, and Alan L. Selman (Eds.). Springer, 218--240. https://doi.org/10.1007/11685654_10Google Scholar
- Alexander Domahidi, Eric Chu, and Stephen P. Boyd. 2013. ECOS: An SOCP solver for embedded systems. In European Control Conference, ECC 2013, Zurich, Switzerland, July 17--19, 2013. IEEE, 3071--3076. http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=6669541Google ScholarCross Ref
- Gene Fliedner. 1999. An investigation of aggregate variable time series forecast strategies with specific subaggregate time series statistical correlation. Computers & operations research, Vol. 26, 10--11 (1999), 1133--1149.Google Scholar
- L. R. Ford and D. R. Fulkerson. 1987. Maximal Flow Through a Network. Birkhäuser Boston, Boston, MA, 243--248. https://doi.org/10.1007/978-0-8176-4842-8_15Google Scholar
- Rob J Hyndman, Roman A Ahmed, George Athanasopoulos, and Han Lin Shang. 2011. Optimal combination forecasts for hierarchical time series. Computational statistics & data analysis, Vol. 55, 9 (2011), 2579--2589.Google Scholar
- Rob J Hyndman, Alan J Lee, and Earo Wang. 2016. Fast computation of reconciled forecasts for hierarchical and grouped time series. Computational statistics & data analysis, Vol. 97 (2016), 16--32.Google Scholar
- Hoel Kervadec, Jose Dolz, Jing Yuan, Christian Desrosiers, Eric Granger, and Ismail Ben Ayed. 2020. Constrained Deep Networks: Lagrangian Optimization via Log-Barrier Extensions. arxiv: 1904.04205 [cs.CV]Google Scholar
- Diederik P. Kingma and Jimmy Ba. 2017. Adam: A Method for Stochastic Optimization. arxiv: 1412.6980 [cs.LG]Google Scholar
- Helmut Lütkepohl. 2011. Forecasting aggregated time series variables: A survey. OECD Journal: Journal of Business Cycle Measurement and Analysis, Vol. 2010, 2 (2011), 1--26.Google ScholarCross Ref
- Frank Nielsen and Richard Nock. 2010. Entropies and cross-entropies of exponential families. In Proceedings of the International Conference on Image Processing, ICIP 2010, September 26--29, Hong Kong, China. IEEE, 3621--3624. https://doi.org/10.1109/ICIP.2010.5652054Google ScholarCross Ref
- B. O'Donoghue, E. Chu, N. Parikh, and S. Boyd. 2016. Conic Optimization via Operator Splitting and Homogeneous Self-Dual Embedding. Journal of Optimization Theory and Applications, Vol. 169, 3 (June 2016), 1042--1068. http://stanford.edu/boyd/papers/scs.htmlGoogle Scholar
- B. O'Donoghue, E. Chu, N. Parikh, and S. Boyd. 2019. SCS: Splitting Conic Solver, version 2.1.2. https://github.com/cvxgrp/scs.Google Scholar
- Boris N. Oreshkin, Dmitri Carpov, Nicolas Chapados, and Yoshua Bengio. 2019. N-BEATS: Neural basis expansion analysis for interpretable time series forecasting. CoRR, Vol. abs/1905.10437 (2019). arxiv: 1905.10437 http://arxiv.org/abs/1905.10437Google Scholar
- Mijung Park and Marcel Nassar. 2014. Variational Bayesian inference for forecasting hierarchical time series. In ICML Workshop. Citeseer.Google Scholar
- Syama Sundar Rangapuram, Matthias W Seeger, Jan Gasthaus, Lorenzo Stella, Yuyang Wang, and Tim Januschowski. 2018. Deep State Space Models for Time Series Forecasting. In Advances in Neural Information Processing Systems 31, S Bengio, H Wallach, H Larochelle, K Grauman, N Cesa-Bianchi, and R Garnett (Eds.). Curran Associates, Inc., 7785--7794.Google Scholar
- David Salinas, Valentin Flunkert, and Jan Gasthaus. 2017. DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks. (April 2017). arxiv: 1704.04110 [cs.AI]Google Scholar
- Rene C. Schaub. 2009. Stochastic Programming Soluctions To Supply Chain Management. Ph.,D. Dissertation. Stanford University. https://web.stanford.edu/group/SOL/dissertations/ReneSchaubThesis.pdfGoogle Scholar
- S.P. Sethi, H. Yan, and H. Zhang. 2006. Inventory and Supply Chain Management with Forecast Updates. Springer US. 2005046552 https://books.google.co.in/books?id=q_53xOvYl1wCGoogle Scholar
- Slawek Smyl. 2020. A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting. International Journal of Forecasting, Vol. 36, 1 (2020), 75--85. https://doi.org/10.1016/j.ijforecast.2019.03.017 M4 Competition.Google ScholarCross Ref
- Richard Stone, David G Champernowne, and James E Meade. 1942. The precision of national income estimates. The Review of Economic Studies, Vol. 9, 2 (1942), 111--125.Google ScholarCross Ref
- Tim Van Erven and Jairo Cugliari. 2015. Game-theoretically optimal reconciliation of contemporaneous hierarchical time series forecasts. In Modeling and stochastic learning for forecasting in high dimensions. Springer, 297--317.Google Scholar
- Shanika L Wickramasuriya, George Athanasopoulos, and Rob J Hyndman. 2019. Optimal forecast reconciliation for hierarchical and grouped time series through trace minimization. J. Amer. Statist. Assoc., Vol. 114, 526 (2019), 804--819.Google ScholarCross Ref
- Guoqiang Peter Zhang. 2003. Time series forecasting using a hybrid ARIMA and neural network model. Neurocomputing, Vol. 50 (2003), 159--175. https://doi.org/10.1016/S0925-2312(01)00702-0Google ScholarCross Ref
Index Terms
- Network Aware Forecasting for eCommerce Supply Planning
Recommendations
SPC forecasting system to mitigate the bullwhip effect and inventory variance in supply chains
We study the impact of forecasting on the bullwhip effect and inventory variance.A forecasting system (SPC-FS) based on control charts is presented and evaluated.Simulation is adopted to evaluate SPC-FS in a supply chain employs OUT policy.SPC-FS leads ...
Collaborative planning, forecasting and replenishment: demand planning in supply chain management
Recognising the value of information sharing among supply chain partners, a growing number of firms have expressed keen interest in jointly forecasting customer demand and co-managing business functions. In particular, such interest sparked the rapid ...
Fuzzy forecasting applications on supply chains
Demand forecasting; which sound basis for decision making process, is among the key activities that directly affect the supply chain performance. As the demand pattern varies from system to system, determination of the appropriate forecasting model that ...
Comments