Intermittent Demand Forecasting: Context, Methods and Applications

ISBN-13: 9781119976080 / Angielski / Twarda / 2018 / 400 str.

John Boylan; Aris Syntetos

Opis

Covers the entire breadth of work in intermittent demand forecasting and the very latest research findings in the following topics: time series (parametric) methods bootstrapping, both parametric and non-parametric causal models neural networks.

Recenzje

Intermittent demand forecasting may seem like a specialized area but actually is at the center of sustainability efforts to consume less and to waste less. Boylan and Syntetos have done a superb job in showing how improvements in inventory management are pivotal in achieving this. Their book covers both the theory and practice of intermittent demand forecasting and my prediction is that it will fast become the 'bible' of the field.Spyros Makridakis, Professor, University of Nicosia, and Director, Institute for the Future and the Makridakis Open Forecasting Center (MOFC)We have been able to support our clients by adopting many of the ideas discussed in this excellent book, and implementing them in our software. I am sure that these ideas will be equally helpful for other supply chain software vendors and for companies wanting to update and upgrade their capabilities in forecasting and inventory management.Suresh Acharya, VP, Research and Development, Blue YonderAs product variants proliferate and the pace of business quickens, more and more items have intermittent demand. Boylan and Syntetos have long been leaders in extending forecasting and inventory methods to accommodate this new reality. Their book gathers and clarifies decades of research in this area, and explains how practitioners can exploit this knowledge to make their operations more efficient and effective.Thomas R. Willemain, Professor Emeritus, Rensselaer Polytechnic Institute

Spis treści

Preface xixGlossary xxiAbout the Companion Website xxiii1 Economic and Environmental Context 11.1 Introduction 11.2 Economic and Environmental Benefits 31.2.1 After-sales Industry 31.2.2 Defence Sector 41.2.3 Economic Benefits 51.2.4 Environmental Benefits 51.2.5 Summary 61.3 Intermittent Demand Forecasting Software 61.3.1 Early Forecasting Software 61.3.2 Developments in Forecasting Software 61.3.3 Open Source Software 71.3.4 Summary 71.4 About this Book 71.4.1 Optimality and Robustness 71.4.2 Business Context 81.4.3 Structure of the Book 91.4.4 Current and Future Applications 101.4.5 Summary 101.5 Chapter Summary 11Technical Note 112 Inventory Management and Forecasting 132.1 Introduction 132.2 Scheduling and Forecasting 132.2.1 Material Requirements Planning (MRP) 132.2.2 Dependent and Independent Demand Items 142.2.3 Make to Stock 152.2.4 Summary 152.3 Should an Item Be Stocked at All? 152.3.1 Stock/Non-Stock Decision Rules 162.3.2 Historical or Forecasted Demand? 182.3.3 Summary 182.4 Inventory Control Requirements 192.4.1 How Should Stock Records be Maintained? 192.4.2 When are Forecasts Required for Stocking Decisions? 222.4.3 Summary 242.5 Overview of Stock Rules 252.5.1 Continuous Review Systems 252.5.2 Periodic Review Systems 262.5.3 Periodic Review Policies 282.5.4 Variations of the (R, S) Periodic Policy 292.5.5 Summary 302.6 Chapter Summary 30Technical Notes 313 Service Level Measures 333.1 Introduction 333.2 Judgemental Ordering 343.2.1 Rules of Thumb for the Order-Up-To Level 343.2.2 Judgemental Adjustment of Orders 343.2.3 Summary 353.3 Aggregate Financial and Service Targets 353.3.1 Aggregate Financial Targets 363.3.2 Service Level Measures 363.3.3 Relationships Between Service Level Measures 383.3.4 Summary 393.4 Service Measures at SKU Level 393.4.1 Cost Factors 393.4.2 Understanding of Service Level Measures 403.4.3 Potential Service Level Measures 403.4.4 Choice of Service Level Measure 413.4.5 Summary 423.5 Calculating Cycle Service Levels 423.5.1 Distribution of Demand Over One Time Period 433.5.2 Cycle Service Levels Based on All Cycles 443.5.3 Cycle Service Levels Based on Cycles with Demand 453.5.4 Summary 473.6 Calculating Fill Rates 483.6.1 Unit Fill Rates 483.6.2 Fill Rates: Standard Formula 493.6.3 Fill Rates: Sobel's Formula 513.6.4 Summary 533.7 Setting Service Level Targets 533.7.1 Responsibility for Target Setting 533.7.2 Trade-off Between Service and Cost 543.7.3 Setting SKU Level Service Targets 553.7.4 Summary 563.8 Chapter Summary 56Technical Note 574 Demand Distributions 594.1 Introduction 594.2 Estimation of Demand Distributions 604.2.1 Empirical Demand Distributions 604.2.2 Fitted Demand Distributions 624.2.3 Summary 644.3 Criteria for Demand Distributions 644.3.1 Empirical Evidence for Goodness of Fit 644.3.2 Further Criteria 644.3.3 Summary 654.4 Poisson Distribution 654.4.1 Shape of the Poisson Distribution 664.4.2 Summary 674.5 Poisson Demand Distribution 674.5.1 Poisson: A Priori Grounds 674.5.2 Poisson: Ease of Calculation 674.5.3 Poisson: Flexibility 684.5.4 Poisson: Goodness of Fit 694.5.5 Testing for Goodness of Fit 704.5.6 Summary 724.6 Incidence and Occurrence 724.6.1 Demand Incidence 724.6.2 Demand Occurrence 734.6.3 Summary 744.7 Poisson Demand Incidence Distribution 754.7.1 A Priori Grounds 754.7.2 Ease of Calculation 754.7.3 Flexibility 764.7.4 Goodness of Fit 764.7.5 Summary 794.8 Bernoulli Demand Occurrence Distribution 794.8.1 Bernoulli Distribution: A Priori Grounds 794.8.2 Bernoulli Distribution: Ease of Calculation 804.8.3 Bernoulli Distribution: Flexibility 814.8.4 Bernoulli Distribution: Goodness of Fit 814.8.5 Summary 824.9 Chapter Summary 82Technical Notes 835 Compound Demand Distributions 875.1 Introduction 875.2 Compound Poisson Distributions 885.2.1 Compound Poisson: A Priori Grounds 895.2.2 Compound Poisson: Flexibility 895.2.3 Summary 895.3 Stuttering Poisson Distribution 905.3.1 Stuttering Poisson: A Priori Grounds 915.3.2 Stuttering Poisson: Ease of Calculation 915.3.3 Stuttering Poisson: Flexibility 935.3.4 Stuttering Poisson: Goodness of Fit for Demand Sizes 935.3.5 Summary 955.4 Negative Binomial Distribution 965.4.1 Negative Binomial: A Priori Grounds 965.4.2 Negative Binomial: Ease of Calculation 965.4.3 Negative Binomial: Flexibility 975.4.4 Negative Binomial: Goodness of Fit 985.4.5 Summary 995.5 Compound Bernoulli Distributions 1005.5.1 Compound Bernoulli: A Priori Grounds 1005.5.2 Compound Bernoulli: Ease of Calculation 1005.5.3 Compound Bernoulli: Flexibility 1005.5.4 Compound Bernoulli: Goodness of Fit 1015.5.5 Summary 1015.6 Compound Erlang Distributions 1015.6.1 Compound Erlang Distributions: A Priori Grounds 1035.6.2 Compound Erlang Distributions: Ease of Calculation 1045.6.3 Compound Erlang-2: Flexibility 1045.6.4 Compound Erlang-2: Goodness of Fit 1045.6.5 Summary 1055.7 Differing Time Units 1055.7.1 Poisson Distribution 1065.7.2 Compound Poisson Distribution 1065.7.3 Compound Bernoulli and Compound Erlang Distributions 1075.7.4 Normal Distribution 1085.7.5 Summary 1105.8 Chapter Summary 110Technical Notes 1116 Forecasting Mean Demand 1176.1 Introduction 1176.2 Demand Assumptions 1186.2.1 Elements of Intermittent Demand 1196.2.2 Demand Models 1196.2.3 An Intermittent Demand Model 1206.2.4 Summary 1216.3 Single Exponential Smoothing (SES) 1216.3.1 SES as an Error-correction Mechanism 1226.3.2 SES as aWeighted Average of Previous Observations 1226.3.3 Practical Considerations 1256.3.4 Summary 1266.4 Croston's Critique of SES 1266.4.1 Bias After Demand Occurring Periods 1266.4.2 Magnitude of Bias After Demand Occurring Periods 1286.4.3 Bias After Review Intervals with Demands 1286.4.4 Summary 1296.5 Croston's Method 1296.5.1 Method Specification 1296.5.2 Method Application 1306.5.3 Summary 1316.6 Critique of Croston's Method 1326.6.1 Bias of Size-interval Approaches 1326.6.2 Inversion Bias 1326.6.3 Quantification of Bias 1336.6.4 Summary 1346.7 Syntetos-Boylan Approximation 1346.7.1 Practical Application 1346.7.2 Framework for Correction Factors 1356.7.3 Initialisation and Optimisation 1356.7.4 Summary 1386.8 Aggregation for Intermittent Demand 1386.8.1 Temporal Aggregation 1386.8.2 Cross-sectional Aggregation 1416.8.3 Summary 1426.9 Empirical Studies 1436.9.1 Single Series, Single Period Approaches 1436.9.2 Single Series, Multiple Period Approaches 1446.9.3 Summary 1456.10 Chapter Summary 145Technical Notes 1467 Forecasting the Variance of Demand and Forecast Error 1517.1 Introduction 1517.2 Mean Known, Variance Unknown 1517.2.1 Mean Demand Unchanging Through Time 1527.2.2 Relating Variance Over One Period to Variance Over the Protection Interval 1527.2.3 Summary 1537.3 Mean Unknown, Variance Unknown 1537.3.1 Mean and Variance Unchanging Through Time 1547.3.2 Mean or Variance Changing Through Time 1557.3.3 Relating Variance Over One Period to Variance Over the Protection Interval 1567.3.4 Direct Approach to Estimating Variance of Forecast Error Over the Protection Interval 1587.3.5 Implementing the Direct Approach to Estimating Variance Over the Protection Interval 1607.3.6 Summary 1607.4 Lead Time Variability 1617.4.1 Consequences of Recognising Lead Time Variance 1617.4.2 Variance of Demand Over a Variable Lead Time (Known Mean Demand) 1627.4.3 Variance of Demand Over a Variable Lead Time (Unknown Mean Demand) 1637.4.4 Distribution of Demand Over a Variable Lead Time 1647.4.5 Summary 1657.5 Chapter Summary 165Technical Notes 1668 Inventory Settings 1698.1 Introduction 1698.2 Normal Demand 1708.2.1 Order-up-to Levels for Four Scenarios 1708.2.2 Scenario 1: Mean and Standard Deviation Known 1708.2.3 Scenario 2: Mean Demand Unknown Standard Deviation Known 1728.2.4 Scenario 3: Mean Demand Known Standard Deviation Unknown 1758.2.5 Scenario 4: Mean and Standard Deviation Unknown 1768.2.6 Summary 1778.3 Poisson Demand 1778.3.1 Cycle Service Level System when the Mean Demand is Known 1778.3.2 Fill Rate System when the Mean Demand is Known 1788.3.3 Poisson OUT Level when the Mean Demand is Unknown 1798.3.4 Summary 1818.4 Compound Poisson Demand 1818.4.1 Stuttering Poisson OUT Level when the Parameters are Known 1818.4.2 Negative Binomial OUT Levels when the Parameters are Known 1838.4.3 Stuttering Poisson and Negative Binomial OUT Levels when the Parameters are Unknown 1838.4.4 Summary 1848.5 Variable Lead Times 1848.5.1 Empirical Lead Time Distributions 1848.5.2 Summary 1858.6 Chapter Summary 185Technical Notes 1869 Accuracy and Its Implications 1939.1 Introduction 1939.2 Forecast Evaluation 1949.2.1 Only One Step Ahead? 1949.2.2 All Points in Time? 1949.2.3 Summary 1959.3 Error Measures in Common Usage 1959.3.1 Popular Forecast Error Measures 1959.3.2 Calculation of Forecast Errors 1979.3.3 Mean Error 1979.3.4 Mean Square Error 1989.3.5 Mean Absolute Error 1989.3.6 Mean Absolute Percentage Error (MAPE) 1989.3.7 100% Minus MAPE 1999.3.8 Forecast Value Added 1999.3.9 Summary 2009.4 Criteria for Error Measures 2009.4.1 General Criteria 2009.4.2 Additional Criteria for Intermittence 2019.4.3 Summary 2019.5 Mean Absolute Percentage Error and its Variants 2019.5.1 Problems with the Mean Absolute Percentage Error 2029.5.2 Mean Absolute Percentage Error from Forecast 2029.5.3 Symmetric Mean Absolute Percentage Error 2039.5.4 MAPEFF and sMAPE for Intermittent Demand 2049.5.5 Summary 2059.6 Measures Based on the Mean Absolute Error 2059.6.1 MAE: Mean Ratio 2059.6.2 Mean Absolute Scaled Error 2069.6.3 Measures Based on Absolute Errors 2079.6.4 Summary 2089.7 Measures Based on the Mean Error 2089.7.1 Desirability of Unbiased Forecasts 2099.7.2 Mean Error 2099.7.3 Mean Percentage Error 2109.7.4 Scaled Bias Measures 2109.7.5 Summary 2119.8 Measures Based on the Mean Square Error 2119.8.1 Scaled Mean Square Error 2129.8.2 Relative Root Mean Square Error 2129.8.3 Percentage Best 2139.8.4 Summary 2139.9 Accuracy of Predictive Distributions 2149.9.1 Measuring Predictive Distribution Accuracy 2149.9.2 Probability Integral Transform for Continuous Data 2159.9.3 Probability Integral Transform for Discrete Data 2159.9.4 Summary 2179.10 Accuracy Implication Measures 2189.10.1 Simulation Outline 2189.10.2 Forecasting Details 2189.10.3 Simulation Details 2199.10.4 Comparison of Simulation Results 2209.10.5 Summary 2219.11 Chapter Summary 221Technical Notes 22110 Judgement, Bias, and Mean Square Error 22510.1 Introduction 22510.2 Judgemental Forecasting 22510.2.1 Evidence on Prevalence of Judgemental Forecasting 22610.2.2 Judgemental Biases 22610.2.3 Effectiveness of Judgemental Forecasts: Evidence for Non-intermittent Items 22910.2.4 Effectiveness of Judgemental Forecasts: Evidence for Intermittent Items 23010.2.5 Summary 23110.3 Forecast Bias 23210.3.1 Monitoring and Detection of Bias 23210.3.2 Bias as an Expectation of a Random Variable 23410.3.3 Response to Different Causes of Bias 23510.3.4 Summary 23610.4 The Components of Mean Square Error 23610.4.1 Calculation of Mean Square Error 23610.4.2 Decomposition of Expected Squared Errors 23610.4.3 Decomposition of Expected Squared Errors for Independent Demand 23810.4.4 Summary 23910.5 Chapter Summary 240Technical Notes 24011 Classification Methods 24311.1 Introduction 24311.2 Classification Schemes 24411.2.1 The Purpose of Classification 24411.2.2 Classification Criteria 24511.2.3 Summary 24511.3 ABC Classification 24611.3.1 Pareto Principle 24611.3.2 Service Criticality 24611.3.3 ABC Classification and Forecasting 24711.3.4 Summary 24811.4 Extensions to the ABC Classification 24811.4.1 Composite Criterion Approach 24911.4.2 Multi-criteria Approaches 25011.4.3 Classification for Spare Parts 25011.4.4 Summary 25111.5 Conceptual Clarifications 25111.5.1 Definition of Non-normal Demand Patterns 25111.5.2 Conceptual Framework 25211.5.3 Summary 25311.6 Classification Based on Demand Sources 25411.6.1 Demand Generation 25411.6.2 A Qualitative Classification Approach 25411.6.3 Summary 25511.7 Forecasting-based Classifications 25511.7.1 Forecasting and Generalisation 25611.7.2 Classification Solutions 25711.7.3 Summary 25811.8 Chapter Summary 259Technical Notes 26012 Maintenance and Obsolescence 26312.1 Introduction 26312.2 Maintenance Contexts 26412.2.1 Summary 26512.3 Causal Forecasting 26512.3.1 Causal Forecasting for Maintenance Management 26612.3.2 Summary 26812.4 Time Series Methods 26812.4.1 Forecasting in the Presence of Obsolescence 26912.4.2 Forecasting with Granular Maintenance Information 27212.4.3 Summary 27312.5 Forecasting in Context 27312.6 Chapter Summary 275Technical Notes 27613 Non-parametric Methods 27913.1 Introduction 27913.2 Empirical Distribution Functions 28013.2.1 Assumptions 28113.2.2 Length of History 28113.2.3 Summary 28213.3 Non-overlapping and Overlapping Blocks 28213.3.1 Differences Between the Two Methods 28213.3.2 Methods and Assumptions 28413.3.3 Practical Considerations 28413.3.4 Performance of Non-overlapping Blocks Method 28513.3.5 Performance of Overlapping Blocks Method 28513.3.6 Summary 28613.4 Comparison of Approaches 28613.4.1 Time Series Characteristics Favouring Overlapping Blocks 28613.4.2 Empirical Evidence on Overlapping Blocks 28713.4.3 Summary 28913.5 Resampling Methods 28913.5.1 Simple Bootstrapping 28913.5.2 Bootstrapping Demand Sizes and Intervals 29013.5.3 VZ Bootstrap and the Syntetos-Boylan Approximation 29213.5.4 Extension of Methods to Variable Lead Times 29313.5.5 Resampling Immediately After Demand Occurrence 29313.5.6 Summary 29413.6 Limitations of Simple Bootstrapping 29413.6.1 Autocorrelated Demand 29413.6.2 Previously Unobserved Demand Values 29513.6.3 Summary 29613.7 Extensions to Simple Bootstrapping 29613.7.1 Discrete-time Markov Chains 29613.7.2 Extension to Simple Bootstrapping Using Markov Chains 29713.7.3 Jittering 29913.7.4 Limitations of Jittering 30013.7.5 Further Developments 30013.7.6 Empirical Evidence on Bootstrapping Methods 30013.7.7 Summary 30213.8 Chapter Summary 302Technical Notes 30314 Model-based Methods 30514.1 Introduction 30514.2 Models and Methods 30514.2.1 A Simple Model for Single Exponential Smoothing 30614.2.2 Critique ofWeighted Least Squares 30714.2.3 ARIMA Models 30714.2.4 The ARIMA(0,1,1) Model and SES 30814.2.5 Summary 30914.3 Integer Autoregressive Moving Average (INARMA) Models 30914.3.1 Integer Autoregressive Model of Order One, INAR(1) 31014.3.2 Integer Moving Average Model of Order One, INMA(1) 31214.3.3 Mixed Integer Autoregressive Moving Average Models 31214.3.4 Summary 31314.4 INARMA Parameter Estimation 31314.4.1 Parameter Estimation for INAR(1) Models 31314.4.2 Parameter Estimation for INMA(1) Models 31414.4.3 Parameter Estimation for INARMA(1,1) Models 31414.4.4 Summary 31514.5 Identification of INARMA Models 31514.5.1 Identification Using Akaike's Information Criterion 31514.5.2 General Models and Model Identification 31614.5.3 Summary 31714.6 Forecasting Using INARMA Models 31714.6.1 Forecasting INAR(1) Mean Demand 31814.6.2 Forecasting INMA(1) Mean Demand 31814.6.3 Forecasting INARMA(1,1) Mean Demand 31914.6.4 Forecasting Using Temporal Aggregation 31914.6.5 Summary 31914.7 Predicting the Whole Demand Distribution 31914.7.1 Protection Interval of One Period 32014.7.2 Protection Interval of More Than One Period 32014.7.3 Summary 32214.8 State Space Models for Intermittence 32214.8.1 Croston's Demand Model 32314.8.2 Proposed State Space Models 32414.8.3 Summary 32514.9 Chapter Summary 325Technical Notes 32515 Software for Intermittent Demand 32915.1 Introduction 32915.2 Taxonomy of Software 33015.2.1 Proprietary Software 33015.2.2 Open Source Software 33215.2.3 Hybrid Solutions 33315.2.4 Summary 33315.3 Framework for Software Evaluation 33315.3.1 Key Aspects of Software Evaluation 33415.3.2 Additional Criteria 33515.3.3 Summary 33615.4 Software Features and Their Availability 33615.4.1 Software Features for Intermittent Demand 33615.4.2 Availability of Software Features 33715.4.3 Summary 33815.5 Training 33915.5.1 Summary 34015.6 Forecast Support Systems 34015.6.1 Summary 34115.7 Alternative Perspectives 34115.7.1 Bayesian Methods 34215.7.2 Neural Networks 34215.7.3 Summary 34315.8 Way Forward 34315.9 Chapter Summary 345Technical Note 345References 347Author Index 365Subject Index 367

O autorze

John E. Boylan is Professor of Business Analytics at Lancaster University, an Editor-in-Chief of the Journal of the Operational Research Society, and President of the International Society for Inventory Research.Aris A. Syntetos is Professor of Operational Research and Operations Management at Cardiff University, an Editor-in-Chief of the IMA Journal of Management Mathematics, and Director of the International Institute of Forecasters.