Article

Marrying math and mind: towards production planning systems for industry 5.0

Taylor & Francis
International Journal of Production Research
Authors:
To read the full-text of this research, you can request a copy directly from the authors.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Many companies, despite there being opportunities for automation in production and logistics (P&L) systems, still rely on human workers due to their cognitive and motor skills. Taking Human Factor (HF) aspects into consideration when making P&L system design and management decisions is therefore important, an ignorance of HF potentially resulting in operator fatigue, discomfort, subsequent injuries and negative consequences for operator performance and the P&L system. A review of the literature shows that the majority of studies that take HF into consideration focus either on designing the workplace or on operation planning activities. There is also still a gap in the literature. Little has been published on P&L systems that incorporate HF and that combine different levels of short-term operational policy decisions (e.g. job allocation) and long-term system characteristic decisions (e.g. layout design). Current state-of-the-art frameworks that support the design and management of P&L systems and that take HF into consideration rarely consider different decision levels. This study proposes a new framework that incorporates HF into P&L systems by combining different levels of decisions to improve performance, quality, and well-being.
Article
Full-text available
Contemporary manufacturing companies are navigating industrial digitalization anticipating increased production efficiency and competitiveness in a volatile environment. This study focuses on the implementation processes of digital tools for production planning and control (PPC), i.e., advanced planning and scheduling (APS) software, in relation to the application of analog planning with physical flow boards. Digital tools can support understanding the consequences of production changes and variations, hence facilitating adaptable and resilient manufacturing. However, technological changes can be daunting, and effective implementations require dynamic capabilities to remain competitive in elusive environments. The aim is to study the implementation processes of an APS software to understand the requirements of fruitfully moving from analog planning to next-generation digital tools for decision support in PPC. The paper presents an explorative case study, at a manufacturing company within the energy sector. The interview study took place over 9 months during 2020–2021, investigating current and retrospective aspects of the case across 2019–2021. The case study comprises 17 in-depth interviews with a range of company employees, e.g., logistics managers and functions responsible for digitalization development. The results highlight the challenges of implementing and especially trusting digital tools for PPC. To realize the value of digital tools for PPC, it is argued that it is imperative to simultaneously apply a human-centric perspective in decision making to ensure trustworthy, sustainable, and resilient human-data-technology nexus implementations towards smart manufacturing.
Article
Full-text available
Past research has found that people treat advice differently depending on its source. In many cases, people seem to prefer human advice to algorithms, but in others, there is a reversal, and people seem to prefer algorithmic advice. Across two studies, we examine the persuasiveness of, and judges' preferences for, advice from different sources when forecasting geopolitical events. We find that judges report domain‐specific preferences, preferring human advice in the domain of politics and algorithmic advice in the domain of economics. In Study 2, participants report a preference for hybrid advice, that combines human and algorithmic sources, to either one on it's own regardless of domain. More importantly, we find that these preferences did not affect persuasiveness of advice from these different sources, regardless of domain. Judges were primarily sensitive to quantitative features pertaining to the similarity between their initial beliefs and the advice they were offered, such as the distance between them and the relative advisor confidence, when deciding whether to revise their initial beliefs in light of advice, rather than the source that generated the advice.
Article
Full-text available
There is empirical evidence that decision makers show negative behaviours towards algorithmic advice compared to human advice, termed as algorithm aversion. Taking a trust theoretical perspective, this study broadens the quite monolithic view on behaviour to its cognitive antecedent: cognitive trust, i.e. trusting beliefs and trusting intentions. We examine initial trust (cognitive trust and behaviour) as well as its development after performance feedback by conducting an online experiment that asked participants to forecast the expected demand for a product. Advice accuracy was manipulated by ± 5 % relative to the participant’s initial forecasting accuracy determined in a pre-test. Results show that initial behaviour towards algorithmic advice is not influenced by cognitive trust. Furthermore, the decision maker’s initial forecasting accuracy indicates a threshold between near-perfect and bad advice. When advice accuracy is at this threshold, we observe behavioural algorithm appreciation, particularly due to higher trusting integrity beliefs in algorithmic advice.
Article
Full-text available
Through a large-scale online field experiment, we provide new empirical evidence for the presence of the anchoring bias in people’s judgement due to irrational reliance on a piece of information that they are initially given. The comparison of the anchoring stimuli and respective responses across different tasks reveals a positive, yet complex relationship between the anchors and the bias in participants’ predictions of the outcomes of events in the future. Participants in the treatment group were equally susceptible to the anchors regardless of their level of engagement, previous performance, or gender. Given the strong and ubiquitous influence of anchors quantified here, we should take great care to closely monitor and regulate the distribution of information online to facilitate less biased decision making.
Article
Full-text available
Actionable Cognitive Twins are the next generation Digital Twins enhanced with cognitive capabilities through a knowledge graph and artificial intelligence models that provide insights and decision-making options to the users. The knowledge graph describes the domain-specific knowledge regarding entities and interrelationships related to a manufacturing setting. It also contains information on possible decision-making options that can assist decision-makers, such as planners or logisticians. This paper proposes a knowledge graph modelling approach to construct actionable cognitive twins for capturing specific knowledge related to production planning and demand forecasting in a manufacturing plant. The knowledge graph provides semantic descriptions and contextualisation of the production lines and processes, including data identification and simulation or artificial intelligence algorithms and forecasts used to support them. Such semantics provide ground for inferencing, relating different knowledge types: creative, deductive, definitional, and inductive. To develop the knowledge graph models for describing the use case thoroughly, systems thinking approach is proposed to design and verify the ontology, develop a knowledge graph and build an actionable cognitive twin. Finally, we evaluate our approach in two use cases developed for a European original equipment manufacturer related to the automotive industry as part of the European Horizon 2020 project FACTLOG.
Article
Full-text available
Forecasts serve as the basis for a wide range of managerial decisions. With the potential of new data sources and new techniques for data analysis, human forecasters are increasingly interacting with algorithms. Although algorithms can show better forecasting performance than humans, forecasters do not always accept these algorithms and instead show aversion to them. Algorithm aversion has become a widely known phenomenon. Drawing on the seminal study of Dietvorst et al. (J Exp Psychol Gen 144(1):114–126, 2015), we extend the evidence on algorithm aversion by introducing three environmental variables from the management accounting literature. We argue that time pressure, “do your best” goals, and forecasters’ data input decision rights on the algorithms input mitigate algorithm aversion. To test our hypotheses, we conducted an experimental study with 1,840 participants overall. We found support for our hypothesis that time pressure mitigates algorithm aversion. We found evidence that the mitigation effect is based on forecasters’ loss of confidence in their own forecast when they are under time pressure. We found no support for our hypothesis on “do your best” goals or forecasters’ data input decision rights.
Article
Full-text available
We analyze the lead time syndrome (LTS), which describes a positive feedback loop between lead time updating and increasing order quantities. While previous studies have focused on theoretical analysis and anecdotal evidence, we examine the impact of human decision making with the help of controlled laboratory studies using a novel experimental design of a simple two tier supply chain. We show that the theoretically postulated phenomenon of the LTS can be reproduced in the lab including human decision makers. We find that the dynamic (load dependent) lead time in the supply chain serves as coordination mechanism among retailers, i.e. experimental participants, but that this mechanism can be distorted by less reliable lead times. Comparing the experimental results to an optimization model in a rolling horizon setting, we further argue that human behavior initiates the vicious cycle of the LTS in our experiments. Finally, we find individual-level evidence for behavioral decision-making, in particular hedging behavior, in our experiment, due to various forms of uncertainty. We discuss the implications of these results in the light of supply chain and production planning systems.
Article
Full-text available
Numerous organizations are striving to maximize the profit of their businesses by the effective implementation of competitive advantages including cost reduction, quick delivery, and unique high-quality products. Effective production-scheduling techniques are methods that many firms use to attain these competitive advantages. Implementing scheduling techniques in high-mix low-volume (HMLV) manufacturing industries, especially in Industry 4.0 environments, remains a challenge, as the properties of both parts and processes are dynamically changing. As a reaction to these challenges in HMLV Industry 4.0 manufacturing, a newly advanced and effective real-time production-scheduling decision-support system model was developed. The developed model was implemented with the use of robotic process automation (RPA), and it comprises a hybrid of different advanced scheduling techniques obtained as the result of analytical-hierarchy-process (AHP) analysis. The aim of this research was to develop a method to minimize the total production process time (total make span) by considering the results of risk analysis of HMLV manufacturing in Industry 4.0 environments. The new method is the combination of multi-broker (MB) optimization and a genetics algorithm (GA) that uses general key process indicators (KPIs) that are easy to measure in any kind of production. The new MB–GA method is compatible with industry 4.0 environments, so it is easy to implement. Furthermore, MB–GA deals with potential risk during production, so it can provide more accurate results. On the basis of survey results, 16% of the asked companies could easily use the new scheduling method, and 43.2% of the companies could use it after a little modification of production.
Article
Full-text available
Background: Deep learning offers considerable promise for medical diagnostics. We aimed to evaluate the diagnostic accuracy of deep learning algorithms versus health-care professionals in classifying diseases using medical imaging. Methods: In this systematic review and meta-analysis, we searched Ovid-MEDLINE, Embase, Science Citation Index, and Conference Proceedings Citation Index for studies published from Jan 1, 2012, to June 6, 2019. Studies comparing the diagnostic performance of deep learning models and health-care professionals based on medical imaging, for any disease, were included. We excluded studies that used medical waveform data graphics material or investigated the accuracy of image segmentation rather than disease classification. We extracted binary diagnostic accuracy data and constructed contingency tables to derive the outcomes of interest: sensitivity and specificity. Studies undertaking an out-of-sample external validation were included in a meta-analysis, using a unified hierarchical model. This study is registered with PROSPERO, CRD42018091176. Findings: Our search identified 31 587 studies, of which 82 (describing 147 patient cohorts) were included. 69 studies provided enough data to construct contingency tables, enabling calculation of test accuracy, with sensitivity ranging from 9·7% to 100·0% (mean 79·1%, SD 0·2) and specificity ranging from 38·9% to 100·0% (mean 88·3%, SD 0·1). An out-of-sample external validation was done in 25 studies, of which 14 made the comparison between deep learning models and health-care professionals in the same sample. Comparison of the performance between health-care professionals in these 14 studies, when restricting the analysis to the contingency table for each study reporting the highest accuracy, found a pooled sensitivity of 87·0% (95% CI 83·0-90·2) for deep learning models and 86·4% (79·9-91·0) for health-care professionals, and a pooled specificity of 92·5% (95% CI 85·1-96·4) for deep learning models and 90·5% (80·6-95·7) for health-care professionals. Interpretation: Our review found the diagnostic performance of deep learning models to be equivalent to that of health-care professionals. However, a major finding of the review is that few studies presented externally validated results or compared the performance of deep learning models and health-care professionals using the same sample. Additionally, poor reporting is prevalent in deep learning studies, which limits reliable interpretation of the reported diagnostic accuracy. New reporting standards that address specific challenges of deep learning could improve future studies, enabling greater confidence in the results of future evaluations of this promising technology. Funding: None.
Article
Full-text available
Copyright: © 2019 INFORMS Behavioral operations is now an established and fundamental research field dedicated to understanding how the behavior of managers, workers, and customers influences operational decisions and outcomes. Just 20 years ago, behavioral research within operations management was relatively small and dispersed. Manufacturing & Service Operations Management has played an integral role in the field’s development since then through its openness to new research questions and methodologies. In this article, we reflect on the research goals pursued in behavioral operations as well as the operational contexts explored. We also share thoughts on promising areas for future research.
Article
Full-text available
In an industrial environment, manufacturing systems may be subject to considerable uncertainties which may lead to numerous schedule disturbances. These disturbances prevent the execution of the planned production schedule. ‘groups of permutable jobs’ is one of the most studied methods that deal with this drawback. It constructs a flexible solution characterised by a set of schedules, allowing a human operator to execute, in real-time, the schedule that best fits the state of the shop. However, because of the limited complexity that one human being can handle, he/she needs to cooperate with the machine in order to take efficient decisions. This paper focuses on investigating the human-machine cooperation for planning and scheduling. A new human-machine interface model assisted with a multi-indicator decision support system has been proposed and evaluated. The results show the usefulness/limits of the proposed model and provide insights into the practice of production planning and scheduling. [Received 27 October 2017; Revised 13 December 2017, 7 June 2018; Accepted 8 June 2018].
Article
Full-text available
Cyber-physical systems lead to an emerging quantity and quality of data that can be collected and processed within the whole production process. [1] This data helps to rationalize and optimize the production planning as well as the operative level. In order to handle this amount of complex information, digital support systems are required. Besides assistant systems, crosslinking of data and machines within a company is one of the central aspects of Industrie 4.0. In a round based training concept the connection between manufacturing execution systems and assistant systems will be explained. Furthermore the changes of work caused by decision support systems as well as the functionalities of assistance systems will be included in the new training concept. Assistant systems can be divided into cognitive and physical assistance. [2] In this article we focus on cognitive assistance. It has perceiving functions and the decision support function because there will be an increasing incidence in the future due to a higher amount of data. To support this theory research results from the project SOPHIE regarding digital and human centered decision support systems will be used. [3] Beside the effects of networking in a company for the employees of production and assembly on the shopfloor the new concept also tries to show the changes on the dispositive level and for employees in the middle management and decision makers.
Article
Full-text available
Although evidence-based algorithms consistently outperform human forecasters, people often fail to use them after learning that they are imperfect, a phenomenon known as algorithm aversion. In this paper, we present three studies investigating how to reduce algorithm aversion. In incentivized forecasting tasks, participants chose between using their own forecasts or those of an algorithm that was built by experts. Participants were considerably more likely to choose to use an imperfect algorithm when they could modify its forecasts, and they performed better as a result. Notably, the preference for modifiable algorithms held even when participants were severely restricted in the modifications they could make (Studies 1–3). In fact, our results suggest that participants’ preference for modifiable algorithms was indicative of a desire for some control over the forecasting outcome, and not for a desire for greater control over the forecasting outcome, as participants’ preference for modifiable algorithms was relatively insensitive to the magnitude of the modifications they were able to make (Study 2). Additionally, we found that giving participants the freedom to modify an imperfect algorithm made them feel more satisfied with the forecasting process, more likely to believe that the algorithm was superior, and more likely to choose to use an algorithm to make subsequent forecasts (Study 3). This research suggests that one can reduce algorithm aversion by giving people some control—even a slight amount—over an imperfect algorithm’s forecast. Data, as supplemental material, are available at https://doi.org/10.1287/mnsc.2016.2643 . This paper was accepted by Yuval Rottenstreich, judgment and decision making.
Article
Full-text available
This study investigate different ways of presenting material and information at workstations while using mixed assembly mode with product variants. The experimental set up annotated an assembly line involving 36 subjects. The material presentation factor involved the use of a material rack compared to using an unstructured kit as well as a structured kit. The information presentation factor involved using a text and number instruction compared to a photograph instruction, and situations with and without component variation were considered. Results show that assembly times and workload ratings were lower when using a kit, whereas using a material rack resulted in perceived decreased workflow and increased stress and frustration. Moreover, assembly times and workload ratings were lower when using photographs, whereas using text and numbers increased mental workload. The results could be useful when planning work places and production systems in order to obtain a better workflow and an increased human performance.
Article
As financial institutions increasingly offer robo‐advisors (RAs) as a service, the empirical evidence on their performance, particularly during market downturns, remains limited. We study the impact of using RAs on investment performance during the 2020 financial crisis caused by the COVID‐19 global pandemic. We obtain daily portfolio and transaction data of investors on an online investment platform. Besides making investment decisions by themselves, investors can also leverage an RA system offered by the platform. We match RA users with other investors who did not use the RA with similar investor and portfolio characteristics before the market crash, then compare their portfolio returns after the crash. We find that RA users experienced significantly fewer losses during the market downturn, and this performance advantage came from the RA‐managed assets (rather than assets managed by RA users themselves). We further show that the RA system adjusted its portfolios to hold less risky funds, whereas human investors stayed with their status quo and did not reduce the risk of their portfolios, which partly accounted for the performance discrepancy. Importantly, the RA’s superior performance was not simply a continuation of its performance in a normal market; in fact, we find RA users to have similar portfolio returns as non‐RA investors during normal markets, and the benefits of RA's risk‐reduction trading strategy only manifested during the financial crisis. Our work offers an empirical assessment of RAs’ performance under different market conditions and advances the understanding of algorithmic decision making in financial markets.
Article
The rapid adoption of artificial intelligence (AI) technologies by many organizations has recently raised concerns that AI may eventually replace humans in certain tasks. In fact, when used in collaboration, machines can significantly enhance the complementary strengths of humans. Indeed, because of their immense computing power, machines can perform specific tasks with incredible accuracy. In contrast, human decision makers (DMs) are flexible and adaptive but constrained by their limited cognitive capacity. This paper investigates how machine-based predictions may affect the decision process and outcomes of a human DM. We study the impact of these predictions on decision accuracy, the propensity and nature of decision errors, and the DM’s cognitive efforts. To account for both flexibility and limited cognitive capacity, we model the human decision-making process in a rational inattention framework. In this setup, the machine provides the DM with accurate but sometimes incomplete information at no cognitive cost. We fully characterize the impact of machine input on the human decision process in this framework. We show that machine input always improves the overall accuracy of human decisions but may nonetheless increase the propensity of certain types of errors (such as false positives). The machine can also induce the human to exert more cognitive efforts, although its input is highly accurate. Interestingly, this happens when the DM is most cognitively constrained, for instance, because of time pressure or multitasking. Synthesizing these results, we pinpoint the decision environments in which human-machine collaboration is likely to be most beneficial. This paper was accepted by Jeannette Song, operations management. Supplemental Material: The data files and online appendices are available at https://doi.org/10.1287/mnsc.2023.4744 .
Article
A pre-planned production workflow may need to adapt to instantaneous states of resources, materials, and workpieces. This adaptation poses an additional challenge when multiple custom workpieces are enmeshed in complex processes. This paper presents a novel framework, Cognitive Engine Process Controller (CEPC), which realizes a human-centric manufacturing system by augmenting the human in workflow selection and providing the flexibility in using alternative methods in the workflow. The CEPC streamlines overwhelming information for the human to select the next manufacturing steps via dynamic web ontology language (OWL) ontologies, a conflict checker, and a learner. The CEPC can monitor the completion of each step, determine the effect of a failed step, and suggest a set of suitable steps. Simulation results showed the potential of the CEPC via demonstrations. The CEPC has also been compared with an existing cognitive engine framework for workflow planning.
Article
We study decision behavior in the selection, forecasting, and production for a new product. In a stylized behavioral model and five experiments, we generate new insight into when and why this combination of tasks can lead to overconfidence (specifically, overestimating the demand). We theorize that cognitive limitations lead to noisy interpretations of signal information, which itself is noisy. Because people are statistically naive, they directly use their noisy interpretation of the signal information as their forecast, thereby underaccounting for the uncertainty that underlies it. This process leads to unbiased forecast errors when considering products in isolation, but leads to positively biased forecasts for the products people choose to launch due to a selection effect. We show that this selection-driven overconfidence can be sufficiently problematic that, under certain conditions, choosing the product randomly can actually yield higher profits than when individuals themselves choose the product to launch. We provide mechanism evidence by manipulating the interpretation noise through information complexity—showing that even when the information is equivalent from a Bayesian perspective, more complicated information leads to more noise, which, in turn, leads to more overconfidence in the chosen products. Finally, we leverage this insight to show that getting a second independent forecast for a chosen product can significantly mitigate the overconfidence problem, even when both individuals have the same information. This paper was accepted by Charles Corbett, operations management.
Article
Various advanced systems deploy artificial intelligence (AI) and machine learning (ML) to improve demand forecasting. Supply chain planners need to become familiar with these systems and trust them, considering real-world complexities and challenges the systems are exposed to. However, planners have the opportunity to intervene based on their experience or information that the systems may not capture. In this context, we study planners’ adjustments to AI-generated demand forecasts. We collect a large amount of data from a leading AI provider and a large European retailer. Our dataset contains 30 million forecasts at the SKU-store-day level for 2019, plus variables related to products, weather, and holidays. In our two-phase analysis, we aim to understand the adjustments made by planners and the quality of these adjustments. Within each phase, we first identify the drivers of adjustments and their quality using random forest, a well-known ML algorithm. Next, we investigate the collective effects of the different drivers on the occurrence and the quality of the adjustments using a decision tree approach. We find that product characteristics such as price, freshness, and discounts are important factors when making adjustments. Large positive adjustments occur more frequently but are often inaccurate, while large negative adjustments are generally more accurate but fewer in number. Thus, planners do not contribute to accuracy on average. Our findings provide insights for the better use of human knowledge in judgmental forecasting.
Article
Steel production scheduling is typically accomplished by human expert planners. Hence, instead of fully automated scheduling systems, steel manufacturers prefer auxiliary recommendation algorithms. Through the suggestion of suitable orders, these algorithms assist human expert planners who are tasked with the selection and scheduling of customer orders. However, it is hard to estimate, what degree of complexity these algorithms should have as steel campaign planning generally lacks precise rule-based procedures; in fact, it requires far-reaching domain knowledge as well as intuition that can only be acquired by years of business experience. Here, contrary to developing new algorithms or improving older ones, we introduce a shuffling-aided network method to assess the complexity of the selection patterns established by a human expert. This technique allows us to formalize and represent the tacit knowledge that enters the campaign planning. As a result of the network analysis, we have discovered that the choice of appropriate customer orders for immediate production is primarily determined by the orders’ carbon content (to be precise: the carbon equivalent). Surprisingly, trace elements like manganese, silicon, and titanium have a lesser impact on the selection decision than assumed by the pertinent literature. Our approach can serve as an input to a range of automated decision-support systems, whenever an expert needs to create groups of orders ('production campaigns') that fulfill certain implicit selection criteria.
Article
Research suggests that algorithms—based on artificial intelligence or linear regression models—make better predictions than humans in a wide range of domains. Several studies have examined the degree to which people use algorithms. However, these studies have been mostly cross‐sectional and thus have failed to address the dynamic nature of algorithm use. In the present paper, we examined algorithm use with a novel longitudinal approach outside the lab. Specifically, we conducted two ecological momentary assessment studies in which 401 participants made financial predictions for 18 days in two tasks. Relying on the judge‐advisor system framework, we examined how time interacted with advice source (human vs. algorithm) and advisor accuracy to predict advice taking. Our results showed that when the advice was inaccurate, people tended to use algorithm advice less than human advice across the period studied. Inaccurate algorithms were penalized logarithmically; the effect was initially strong but tended to fade over time. This suggests that first impressions are crucial and produce significant changes in advice taking at the beginning of the interaction, which later tends to stabilize as days go by. Therefore, inaccurate algorithms are more likely to accrue a negative reputation than inaccurate humans, even when having the same level of performance.
Article
Production planning, scheduling and sequencing comprise the core of the manufacturing companies’ performance. The new and changing market demands make manufacturing a challenge because companies must produce by using the minimum possible number of resources to provide high-quality products and to respond quickly to market demands. Thus the need for efficient production planning, scheduling and sequencing has become a very important research area for companies and researchers in recent decades. We evaluated the current state of such research with a holistic framework that comprised the plans aggregation and disaggregation levels, the modelling approaches to represent the different types of plans and their characteristics, the solution approaches with the adopted algorithms, the application areas, the intra- and inter-enterprise levels of integration, the sizes of the datasets used to validate the models and algorithms, the development tools, and the quality of the solutions obtained in relation to the problems’ data size. The systematic literature review is arranged within the framework and grouped around different types of plans, including production planning, scheduling and sequencing, and their combinations. Finally, some gaps in the related research are identified and future research opportunities are proposed.
Chapter
Digitalization enables the full potential of Artificial Intelligence for the first time. This study deals with demand forecasting as a representation of supply chain planning. Statistical and judgmental approaches constitute the state-of-the-art in methodology, but also present drawbacks such as human mental capacity constraints or data biases. Teaming of humans and AI promises synergies and better solutions, but challenging questions on how to organize collaborative tasks remain. Information asymmetry states an unsolved issue, as digitalization is going to take more time to be holistically established. Deploying a behavior analysis of an industrial case study, this paper investigates the impact of two different forms of interactions on the forecasting performance and ability of human planners to compensate the lack of contextual information included in an AI-based prediction. The results indicate that information asymmetry limits the magnitude of the decision-making anchor provided by the algorithm and affects the accuracy depending on the specific interaction form. Overall, an asymmetric sequential interaction set-up outperforms the other forecasts. Finally, this study states implications and limitations for human-AI collaboration.
Article
A four-year series of subjective probability forecasting tournaments sponsored by the U.S. intelligence community revealed a host of replicable drivers of predictive accuracy, including experimental interventions such as training in probabilistic reasoning, anti‐groupthink teaming, and tracking of talent. Drawing on these data, we propose a Bayesian BIN model (Bias, Information, Noise) for disentangling the underlying processes that enable forecasters and forecasting methods to improve—either by tamping down bias and noise in judgment or by ramping up the efficient extraction of valid information from the environment. The BIN model reveals that noise reduction plays a surprisingly consistent role across all three methods of enhancing performance. We see the BIN method as useful in focusing managerial interventions on what works when and why in a wide range of domains. An R-package called BINtools implements our method and is available on the first author’s personal website. This paper was accepted by Manel Baucells, decision analysis.
Article
Computer-based demand forecasting systems have been widely adopted in supply chain companies, but little research has studied how these systems are actually used in the forecasting process. We report the findings of a case study of demand forecasting in a pharmaceutical company over a 15-year period. At the start of the study, managers believed that they were making extensive use of their forecasting system that was marketed based on the accuracy of its advanced statistical methods. Yet most forecasts were obtained using the system’s facility for judgmentally overriding the automatic statistical forecasts. Carrying out the judgmental interventions involved considerable management effort as part of a sales & operations planning (S&OP) process, yet these often only served to reduce forecast accuracy. This study uses observations of the forecasting process, interviews with participants and data on the accuracy of forecasts to investigate why the managers continued to use non-normative forecasting practices for many years despite the potential economic benefits that could be achieved through change. The reasons for the longevity of these practices are examined both from the perspective of the individual forecaster and the organization as a whole.
Article
Will people use self-driving cars, virtual doctors, and other algorithmic decision-makers if they outperform humans? The answer depends on the uncertainty inherent in the decision domain. We propose that people have diminishing sensitivity to forecasting error and that this preference results in people favoring riskier (and often worse-performing) decision-making methods, such as human judgment, in inherently uncertain domains. In nine studies (N = 4,820), we found that (a) people have diminishing sensitivity to each marginal unit of error that a forecast produces, (b) people are less likely to use the best possible algorithm in decision domains that are more unpredictable, (c) people choose between decision-making methods on the basis of the perceived likelihood of those methods producing a near-perfect answer, and (d) people prefer methods that exhibit higher variance in performance (all else being equal). To the extent that investing, medical decision-making, and other domains are inherently uncertain, people may be unwilling to use even the best possible algorithm in those domains.
Article
Data-driven decision-making (DDD) is rapidly transforming modern operations. The availability of big data, advances in data analytics tools, and rapid gains in processing power enable firms to make decisions based on data rather than intuition. Yet, most firms still allow managers to override decisions from DDD tools, as managers might possess private information not present in the DDD tool. We report on a field-experiment conducted by an automobile replacement parts retailer that examines the profit implications of providing discretionary power to merchants. We find that merchants’ overrides of the DDD tool reduce profitability by 5.77%. However, our analysis over product life cycle (PLC) reveals that merchants increase (decrease) profitability for growth- (mature- & decline-) stage products. This paper was accepted by Charles Corbett, operations management.
Article
The unawareness of production planners about the interaction of key performance indicators (KPIs) in manufacturing systems and misinterpretations of system states often leads to problems when they try to improve them. Exemplary, the lead time syndrome (LTS) represents one of these counterproductive actions. While planners aim to improve due date reliability by planned lead time adjustments, the result is often an aggravation of due date reliability. The underlying reason is that human perception and decision-making process can be biased. Despite its potential to improve the logistic performance, the meaning and the effects of cognitive biases on the decision-making processes in production planning and control were out of scope of recent investigations in the field of production logistics. The aim of this research is to create a starting point to close this research gap by the development of a heuristic framework identifying relevant decision making situations, the potentially active cognitive biases and the potential impact on logistics performance; for this, we combine the research streams of production planning and control (PPC), behavioral supply chain management and psychology.
Chapter
The trend of digitalization has led to disruptive changes in production and supply chain planning, where autonomous machines and artificial intelligence gain competitive advantages. Besides, the satisfaction of customers’ wishes has reached top priority for demand-driven companies. Consequently, companies implement digital applications, for instance neural networks for accurate demand forecasting and optimized decision-making tools, to cope with nervous operational planning activities. Since planning tasks require human-machine interaction to increase performance and efficiency of planning decisions, this analysis focuses on forms of interaction to determine the right level of collaboration. The paper outlines various levels of interaction and analyses the impact of human reactions in the context of an industrial demand planning algorithm use case at Infineon Technologies AG conducting a behavioral experiment. The results show that a variance in the levels of human-machine interaction has influence on human acceptance of algorithms, but further experiments need to be conducted to outline an overall framework.
Conference Paper
We address a relatively under-explored aspect of human-computer interaction: people's abilities to understand the relationship between a machine learning model's stated performance on held-out data and its expected performance post deployment. We conduct large-scale, randomized human-subject experiments to examine whether laypeople's trust in a model, measured in terms of both the frequency with which they revise their predictions to match those of the model and their self-reported levels of trust in the model, varies depending on the model's stated accuracy on held-out data and on its observed accuracy in practice. We find that people's trust in a model is affected by both its stated accuracy and its observed accuracy, and that the effect of stated accuracy can change depending on the observed accuracy. Our work relates to recent research on interpretable machine learning, but moves beyond the typical focus on model internals, exploring a different component of the machine learning pipeline.
Article
We investigate the impact of behavioral ordering on profits under competition. Specifically, we use controlled laboratory experiments to evaluate the differences in profits between a behavioral competitor (where a human places orders), and a management science‐driven competitor (where orders are placed according to one of several plausible policies based on existing literature and managerial practice). Unlike the full‐information game‐theoretic models that assume rational decision‐makers, these policies mimic practical situations by using less information and do not assume that their human competitors make fully rational decisions. Most prior literature focuses on non‐competitive settings, where behaviorally‐biased deviations from optimal order quantities result in small expected profit losses. Under competition, we find that human decision‐makers receive a substantially lower profit than the equilibrium expected profit, even as their competitors receive substantially higher profit. This article is protected by copyright. All rights reserved.
Article
People made forecasts from real data series. The points in the series were un‐trended and independent. Hence, forecasts should have been on the mean value. However, consistent with previous research on forecasting biases, forecasts were too close to the last data point. It appears that forecasters see positive sequential dependence where none exists. In three experiments, we examined this bias in different types of forecasting task: point forecasting, probability density forecasting, and interval forecasting. In all cases, we found that it was greater when the data series were displayed using continuous line graphs than when it was displayed using discrete point graphs. Consistent with arguments made by Zacks and Tversky (Memory and Cognition, 27:1073, 1999), we suggest that people are more likely to group data together and to see patterns in them when those data are presented in a continuous than in a discrete format. These findings have implications for forecasting practice.
Article
Demand forecasts are the lifeblood of supply chains. Academic literature and common industry practices indicate that demand forecasts are often subject to human interventions. Judgmental forecasting or judgmental forecast adjustments can cause both positive and negative repercussions to the rest of the supply chain. This paper provides the first systematic literature review of judgmental forecasting and adjustments focusing on key features that impact various decisions in supply chains. A carefully assembled and shortlisted literature pool is analyzed for systematic mapping of the published works using bibliometric tools. The primary sub streams of research within the broader scope of the field are synthesized from a rigorous keyword cluster analysis and a thorough discussion is presented. Our review concludes by encapsulating the key learnings from four decades of academic research in judgmental forecasting and suggests future research avenues to expand our understanding of the role of humans in demand forecasting and supply chain decision-making.
Article
This paper examines the accuracy of judgmental forecasts of product demand and the quality of subsequent production level decisions under two different conditions: (i) the availability of only time series information on past demand; (ii) the availability of time series information together with scenarios that outline possible prospects for the product in the forthcoming period. An experiment indicated that production level decisions made by participants had a greater deviation from optimality when they also received optimistic and pessimistic scenarios. This resulted from less accurate point forecasts made by these participants. Further analysis suggested that participants focussed on the scenario that was congruent with the position of the latest observation relative to the series mean and discounted the opposing scenario. This led to greater weight being attached to this observation, thereby exacerbating the tendency of judgmental forecasters to see systematic changes in random movements in time series.
Article
Product forecasts are a critical input into sourcing, procurement, production, inventory, logistics, finance and marketing decisions. Numerous quantitative models have been developed and applied to generate and improve product forecasts. The use of human judgement, either solely or in conjunction with quantitative models, has been well researched in the academic literature and is a popular forecasting approach in industry practice. In the context of judgemental forecasting, methods that integrate an expert's judgement into quantitative forecasting models are commonly referred to as “integrating forecasting” methods. This paper presents a systematic review of the literature of judgemental demand forecasting with a focus placed on integrating methods. We explore the role of expert opinion and contextual information and discuss the application of behaviourally informed support systems. We also provide important directions for further research in these areas.
Article
At present, sustainability science is mainly based on conscious information and strongly focused on analytical tools or strategies. Neuroscience has made obvious that human decisions are prepared by the unconsciousness. Intuition plays an important role in early and late stages of learning processes and has a crucial impact on decision-making. Thus, intuitive and unconscious thinking is crucial for management processes in general and production planning processes in the main. However, unconscious knowledge and human behaviour is predominantly neglected in production research. Especially the addressing of human machine interfaces (HMI), human cognitive biases have a crucial impact on decision making processes. Constellation work is based on unconscious knowledge and intuition. Thus, systemic structural constellations are an innovative tool to integrate unconscious knowledge in a research context. In systemic structural constellations specific foci of complex systems, such as a production system, can be simulated and represented through spatial arrangements of persons or symbols. So, the method was used to reveal relevant patterns of relationships, structures, interaction, implicit knowledge, including hidden or underlying dynamics and influences that are relevant to and within a production system to understand how the raised problems in HMI can be better solved. The guiding research question is: How can the use of structural systemic constellations improve decision-making processes in HMI contexts in production environments in order to increase sustainability? Results show sustainability seems to be a matter of consciousness and is closely linked to the bias group not enough meaning. Sustainability and complexity resemble more than being linked by trade-offs. The recognition of human biases can be trained to improve human-machine-interfaces and sustainability. Constellation work contributes to decision theory by supporting effectuation.
Book
Commercial advanced planning and scheduling (APS) systems have been around for about 25 years and have seen widespread adoption in a variety of companies. The promises made by APS system implementations are high, and at the same time, APS projects are complex, costly, and carry a significant risk of failure. There is a great need for guidance on how to successfully implement APS systems, which is why this book was written. In this book, we aim to discuss all facets of APS implementation, from theoretical background to definition, vendor selection, system design and the implementation process. In addition, this book helps readers understand how the underlying concepts were formed, why a concept will work in practice, and when it will not. The contents of this book are based on many years of academic research and APS implementation experience. It draws together theory and practice in production control and explains how theories can be applied to support planning and scheduling processes in practice. This book will help anyone involved in APS implementations to achieve success: human planners generate better plans than before, using the APS.
Article
Existing evidence suggests that managers exhibit a censorship bias: demand beliefs tend to be biased low when lost sales from stockouts are unobservable (censored demand) compared to when they are observable (uncensored demand). We develop a non-constraining, easily-implementable behavioral debias technique to help mitigate this tendency in demand forecasting and inventory decision-making settings. The debiasing technique has individuals record estimates of demand outcomes (REDO): participants explicitly record a self-generated estimate of every demand realization, allowing them to record a different value than the number of sales in periods with stockouts. In doing so, they construct a more representative sample of demand realizations (that differs from the sales sample). In three laboratory experiments with MBA and undergraduate students, this remedy significantly reduces downward bias in demand beliefs under censorship and leads to higher inventory order decisions.
Article
Many decisions can be analyzed and supported by quantitative models. These models tend to be complex psychologically in that they require the elicitation and combination of quantities such as probabilities, utilities, and weights. They may be simplified so that they become more transparent, and lead to increased trust, reflection, and insight. These potential benefits of simplicity should be weighed against its potential costs, notably possible decreases in performance. We review and synthesize research that has used mathematical analyses and computer simulations to investigate if and when simple models perform worse, equal, or better than more complex models. Various research strands have pursued this, but have not reached the same conclusions: Work on frequently repeated decisions as in inference and forecasting—which typically are operational and involve one or a few decision makers—has put forth conditions under which simple models are more accurate than more complex ones, and some researchers have proposed that simple models should be preferred. On the other hand, work on more or less one-off decisions as in preference and multi-criteria analysis—which typically are strategic and involve group decision making and multiple stakeholders—has concluded that simple models can at best approximate satisfactorily the more complex models. We show how these conclusions can be reconciled. Additionally, we discuss the theory available for explaining the relative performance of simple and more complex models. Finally, we present an aid to help determine if a simple model should be used, or not, for a particular type of decision problem.
Conference Paper
One of the important functions of the operation of integrated management information systems, including multi-agent systems, is to properly planning the production. Due to the different production planning strategies (methods) and the company’s limited production capacity, the agents running in the system may generate different versions of the production plans. In other word, agents’ knowledge may differ. The final version may be selected by the system user, however, it should be noted that this is a time-consuming process, and there is a risk of the user choosing the worst version. The better solution is to automatically integrate the agents’ knowledge and to determine one version of the plan presented to the user. The aim of this paper is to develop a consensus algorithm that will allow integrating manufacturing plans generated by different agents, and present one solution (that is very close to these plans, but not necessarily one of them) to user.
Article
Continued usage of new Enterprise Resource Planning (ERP) systems has plagued organizations that intend to maximize long-term benefits from their ERP investments. Leadership behavior is widely regarded as one of the key influences for motivating ERP users toward using the system. This study investigates how direct supervisors’ leadership styles influence ERP users’ motivation to continuously engage with the ERP system. We employed self-determination theory (SDT) and the post-acceptance model of information systems to propose a conceptual model theorizing how transformational and transactional leadership styles affect users’ intrinsic and extrinsic motivation, which in turn impacts ERP continuance intentions through user satisfaction and perceived usefulness. Our research model was empirically examined using data collected from 299 ERP users. Our findings have revealed that transformational leadership motivates ERP users differently than transactional leadership, and that user satisfaction and perceived usefulness are salient predictors of ERP continuance intentions. In addition, our research demonstrates a critical role of direct managers’ leadership styles in the ERP post-implementation phase. Important theoretical contributions and significant implications for practice are discussed.
Article
This work emphasizes on the implementation of soft computing techniques for the production planning of complex manufacturing plants. A hierarchical control structure has been assumed to control and to optimize the chemical process of a complex plant with successful results and at the highest hierarchical level the production planning should meet the fluctuating demands in optimal way. This level integrates the operational and business management requirements where the soft computing technique of Fuzzy Cognitive Maps is proposed to model these tasks. The strategic planning for the hierarchical integrated system realizes the optimal total management at the corporate level. The obtained simulation results prove the applicability of the proposed methodology and the advantages of using soft computing modeling techniques for the sophisticated high level of hierarchical systems.