Conference Paper

Queue Mining – Predicting Delays in Service Processes

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

Abstract

Information systems have been widely adopted to support service processes in various domains, e.g., in the telecommunication, finance, and health sectors. Recently, work on process mining showed how management of these processes, and engineering of supporting systems, can be guided by models extracted from the event logs that are recorded during process operation. In this work, we establish a queueing perspective in operational process mining. We propose to consider queues as first-class citizens and use queueing theory as a basis for queue mining techniques. To demonstrate the value of queue mining, we revisit the specific operational problem of online delay prediction: using event data, we show that queue mining yields accurate online predictions of case delay.

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.

... Many studies have been conducted in order to deal with various prediction tasks such as predicting the remaining processing time [4,63,54,52,53], predicting the outcomes of a process [37,22,67,50], predicting future events [23,63,27], etc (cf. [43,42,58,49,15,19]). An overview of various works in the area of predictive business process monitoring can be found in [38,24]. ...
... The work by [54,55] proposes a technique for predicting the remaining processing time using stochastic petri nets. The works by [58,59,42,49] focus on predicting delays in process execution. In [58,59], the authors use queueing theory to address the problem of delay prediction, while [42] explores the delay prediction in the domain of transport and logistics process. ...
... The works by [58,59,42,49] focus on predicting delays in process execution. In [58,59], the authors use queueing theory to address the problem of delay prediction, while [42] explores the delay prediction in the domain of transport and logistics process. In [28], the authors present an ad-hoc predictive clustering approach for predicting process performance. ...
Preprint
Full-text available
Predictive analysis in business process monitoring aims at forecasting the future information of a running business process. The prediction is typically made based on the model extracted from historical process execution logs (event logs). In practice, different business domains might require different kinds of predictions. Hence, it is important to have a means for properly specifying the desired prediction tasks, and a mechanism to deal with these various prediction tasks. Although there have been many studies in this area, they mostly focus on a specific prediction task. This work introduces a language for specifying the desired prediction tasks, and this language allows us to express various kinds of prediction tasks. This work also presents a mechanism for automatically creating the corresponding prediction model based on the given specification. Differently from previous studies, instead of focusing on a particular prediction task, we present an approach to deal with various prediction tasks based on the given specification of the desired prediction tasks. We also provide an implementation of the approach which is used to conduct experiments using real-life event logs.
... A closely related paper to ours is Shah et al. (2019), where the authors describe results from a simulation study which extends our results, e.g., by quantifying the impact of using future information for LES and considering an alternative Newsvendor-type error criterion. For an empirical investigation of the (superior) performance of the LES announcement with real-life data, see Senderovich et al. (2014 and. In contrast, our data-based study focuses here on settings where the LES announcement performs poorly relative to EA. ...
... In this paper, we supplement the literature by proposing a new correlation-based framework which allows for a broader understanding of performance across different queueing models. Through our framework, we are able to provide a theoretical justification to earlier numerical and empirical observations, e.g., in Thiongane et al. (2016) and Senderovich et al. (2014 and. Importantly, our framework can be useful in practice because estimating correlations is easier than fitting queueing models to data. ...
... From a practical and empirical standpoint, there is some empirical evidence substantiating the good performance of the LES announcement with real-life data in some cases; e.g., see Senderovich et al. (2014), Senderovich et al. (2015), and Gal et al. (2017). In contrast, we begin here by presenting conflicting empirical evidence which illustrates the poor accuracy of the LES announcement in other cases, relative to the static announcement EA; we will return to this empirical evidence in §6.2.1. ...
Article
Service providers often share delay information, in the form of delay announcements, with their customers. In practice, simple delay announcements, such as average waiting times or a weighted average of previously delayed customers, are often used. Our goal in this paper is to gain insight into when such announcements perform well. Specifically, we compare the accuracies of two announcements: (i) a static announcement that does not exploit real-time information about the state of the system and (ii) a dynamic announcement, specifically the last-to-enter-service (LES) announcement, which equals the delay of the last customer to have entered service at the time of the announcement. We propose a novel correlation-based approach that is theoretically appealing because it allows for a comparison of the accuracies of announcements across different queueing models, including multiclass models with a priority service discipline. It is also practically useful because estimating correlations is much easier than fitting an entire queueing model. Using a combination of queueing-theoretic analysis, real-life data analysis, and simulation, we analyze the performance of static and dynamic announcements and derive an appropriate weighted average of the two which we demonstrate has a superior performance using both simulation and data from a call center. This paper was accepted by Vishal Gaur, operations management.
... If the quality is below the threshold, a traditional control-flow-based log split is computed and the respective control-flow operator is added to the CaT (lines 11-12) (see Section II-B). Next, regardless of the applied type of log split (data or control-flow), the obtained sublogs are handled (lines [15][16][17][18][19][20][21][22][23]. That is, we check for the trivial cases of an empty sublog or a sublog with events of a single activity (line 17), and add the respective node to the CaT (line 18). ...
... Specifically, as first step, a control-flow model (e.g., a Petri net) is discovered. In a second step, the model is annotated with decisions [13], [24], resources and time [25], queues [22], and general context information [26]. These approaches improve performance analysis and predictive monitoring using the additional context in the enhancement stage, whereas in our work, we focus on improving the quality of the resulting model at the discovery stage. ...
... The synthetic logs are publicly available at https://github.com/dafna-s/ Inductive-Context-aware-Process-Discovery/tree/master/Synthetic log6 We assume Poisson arrivals, which is a common arrival scheme in the literature[22]. A truncated distribution is a conditional distribution that restricts the domain of a given distribution. ...
... Such delay-history-based predictions can perform remarkably well, for example, in large heavily congested systems with or without customer abandonment, even when customers respond to the announcements [49]. There is also some empirical evidence substantiating their good performance in practice [71,72]. However, they do not perform well in other settings, such as when the system is small or lightly loaded [78,90], or under time-varying conditions [52]. ...
... The recent proliferation of empirical studies, in the context of delay announcements, prompts one to evaluate the alternative methods that are used to address that problem. In broad terms, the literature ranges from analytical work, typically substantiated by simulation-based results ( [8,83], etc.), to empirical work in the context of a well-defined structural model ( [3,81,89], etc.), to work which relies, for the most part, on data-mining meth-ods [7,[70][71][72]. Each body of work is important in its own right, and it is crucial to emphasize the complementarity of those different approaches. ...
... However, data-mining techniques are limited in that they are "black-box" techniques that do not, in general, further our understanding about the dynamics of the system. Recently, the combination of those two frameworks (queueing and data-based) has been advocated in several papers [7,[70][71][72]. Indeed, the delay predictors in those papers are inspired by both queueing-theoretic methods and data-mining techniques and are shown to yield superior performance with real-life data sets. ...
Article
Full-text available
Service providers routinely share information about upcoming waiting times with their customers, through delay announcements. The need to effectively manage the provision of these announcements has led to a substantial growth in the body of literature which is devoted to that topic. In this survey paper, we systematically review the relevant literature, summarize some of its key ideas and findings, describe the main challenges that the different approaches to the problem entail, and formulate research directions that would be interesting to consider in future work.
... (F) [114], [123], [113], [115], [49], [14], [130], [128], [50] Control, Design, and Uncertainty Quantification: ...
... This idea has been explored in a series of recent papers. In [114] Sendrovich, Weidlich, Gal, and Mandelbaum use the developed field of business process mining based on event logs, see [123], for queues. They adapted ideas from this field to queues and developed the method of queue mining. ...
Preprint
Full-text available
We present a broad literature survey of parameter and state estimation for queueing systems. Our approach is based on various inference activities, queueing models, observations schemes, and statistical methods. We categorize these into branches of research that we call estimation paradigms. These include: the classical sampling approach, inverse problems, inference for non-interacting systems, inference with discrete sampling, inference with queueing fundamentals, queue inference engine problems, Bayesian approaches, online prediction, implicit models, and control, design, and uncertainty quantification. For each of these estimation paradigms, we outline the principles and ideas, while surveying key references. We also present various simple numerical experiments. In addition to some key references mentioned here, a periodically-updated comprehensive list of references dealing with parameter and state estimation of queues will be kept in an accompanying annotated bibliography.
... However, when observing a queue, n a and n s are often not fixed and can be dependent on the sequences {A i } and {S i }. Complications may arise, not only due [14,49,50,[113][114][115][116]123,128,130] Control, Design, and Uncertainty Quantification: This paradigm deals with work related to parameter and state estimation where control and design decisions based on inferred values are to be made [8,10,17,48,74,87] to censoring, but also due to the dependency structure of the various quantities. Here are some possibilities: ...
... This idea has been explored in a series of recent papers. In [115] Sendrovich, Weidlich, Gal, and Mandelbaum use the developed field of business process mining based on event logs, see [123], for queues. They adapted ideas from this field to queues and developed the method of queue mining. ...
Article
Full-text available
We present a broad literature survey of parameter and state estimation for queueing systems. Our approach is based on various inference activities, queueing models, observations schemes, and statistical methods. We categorize these into branches of research that we call estimation paradigms. These include: the classical sampling approach, inverse problems, inference for non-interacting systems, inference with discrete sampling, inference with queueing fundamentals, queue inference engine problems, Bayesian approaches, online prediction, implicit models, and control, design, and uncertainty quantification. For each of these estimation paradigms, we outline the principles and ideas, while surveying key references. We also present various simple numerical experiments. In addition to some key references mentioned here, a periodically updated comprehensive list of references dealing with parameter and state estimation of queues will be kept in an accompanying annotated bibliography.
... Various approaches exist for dealing with incomplete data of processes with non-isolated cases that compete for scarce resources. In call-center processes, thoroughly studied in [11], queueing theory models can be used for load predictions under assumptions about distributions of unobserved parameters, such as customer patience duration [12], while assuming high load snapshot principle predictors show better accuracy [13]. For time predictions in congested systems, the required features are extracted using congestion graphs [14] mined using queuing theory. ...
... Sect. 2) approaches the problem of analyzing the performance of systems with shared resources primarily either from the control-flow perspective [17,19,20,10,5] or the resource/queuing perspective [11,12,13,14], leading to information loss about the other perspective. In the following, we show how to conceptualize the problem from both perspectives at once using synchronous proclets [6] extended with a few concepts of coloured Petri Nets [7]. ...
Preprint
Full-text available
To identify the causes of performance problems or to predict process behavior, it is essential to have correct and complete event data. This is particularly important for distributed systems with shared resources, e.g., one case can block another case competing for the same machine, leading to inter-case dependencies in performance. However, due to a variety of reasons, real-life systems often record only a subset of all events taking place. For example, to reduce costs, the number of sensors is minimized or parts of the system are not connected. To understand and analyze the behavior of processes with shared resources, we aim to reconstruct bounds for timestamps of events that must have happened but were not recorded. We present a novel approach that decomposes system runs into entity traces of cases and resources that may need to synchronize in the presence of many-to-many relationships. Such relationships occur, for example, in warehouses where packages for N incoming orders are not handled in a single delivery but in M different deliveries. We use linear programming over entity traces to derive the timestamps of unobserved events in an efficient manner. This helps to complete the event logs and facilitates analysis. We focus on material handling systems like baggage handling systems in airports to illustrate our approach. However, the approach can be applied to other settings where recording is incomplete. The ideas have been implemented in ProM and were evaluated using both synthetic and real-life event logs.
... anomaly detection, leftover instance duration prediction, QoS parameter analysis, resource assignment to activities, organisational relations and others. These techniques fall under the umbrella of Operational Process Mining (Senderovich, Weidlich, Gal, & Mandelbaum, 2014c). ...
... For discovering BP behaviour parameters that can be used in the simulation, process mining provides multiple solutions. To name a few of such state of the art methods: Rozinat provides decision rule mining that can be used to predict branching (Rozinat & Van Der Aalst, 2006); Van Dongen et al. in have presented an approach for predicting activity durations; Senderovich et al. apply process mining for discovering resource scheduling protocols (Senderovich et al., 2014c). While there are many possible applications of process mining methods for discovering BP behaviour, their application is limited to discovering parameters that can be used for simulation. ...
... Limitations of the queueing-theoretic analysis have led to recent interest in data-based methods such as machine-learning algorithms and data-mining techniques. Combining process mining and queueing-theoretic results, a technique called queue-mining is introduced in [9] for predicting waiting times in service systems. In [10], the authors propose a new predictor, called Q-Lasso, which combines the Lasso method from statistical learning and fluid models from the queueing theory. ...
... In [10], the authors propose a new predictor, called Q-Lasso, which combines the Lasso method from statistical learning and fluid models from the queueing theory. Similar to [9] and [10], most of the existing works in this area focus on single-value delay predictions and provide no information on the distribution of the delay. A closely related work to this paper is [11], which studies delay distribution prediction in single stage queueing systems using delay history information. ...
Preprint
Full-text available
Ensuring the conformance of a service system's end-to-end delay to service level agreement (SLA) constraints is a challenging task that requires statistical measures beyond the average delay. In this paper, we study the real-time prediction of the end-to-end delay distribution in systems with composite services such as service function chains. In order to have a general framework, we use queueing theory to model service systems, while also adopting a statistical learning approach to avoid the limitations of queueing-theoretic methods such as stationarity assumptions or other approximations that are often used to make the analysis mathematically tractable. Specifically, we use deep mixture density networks (MDN) to predict the end-to-end distribution of the delay given the network's state. As a result, our method is sufficiently general to be applied in different contexts and applications. Our evaluations show a good match between the learned distributions and the simulations, which suggest that the proposed method is a good candidate for providing probabilistic bounds on the end-to-end delay of more complex systems where simulations or theoretical methods are not applicable.
... Their proposed estimators are based on a real-time history of the queue, including last-to-enter-service (LES) information (i.e., the delay experienced by the last customer entering service) and head-of-line (HOL) data (the total delay experienced by the customer currently at the head of the line). These estimators perform well in reality, as was shown in Senderovich et al. (2014). Senderovich et al. (2014) used queue mining techniques to solve the on-line delay prediction problem, validating the theory-based queueing predictors with real data. ...
... These estimators perform well in reality, as was shown in Senderovich et al. (2014). Senderovich et al. (2014) used queue mining techniques to solve the on-line delay prediction problem, validating the theory-based queueing predictors with real data. ...
Article
We investigate the impact of delay announcements on the coordination within hospital networks using a combination of empirical observations and numerical experi- ments. We offer empirical evidence that suggests that patients take delay information into account when choosing emergency service providers and that such information can help increase coordination in the network, leading to improvements in the performance of the network, as measured by emergency department wait times. Our numerical results indi- cate that the level of coordination that can be achieved is limited by the patients’ sensitivity to waiting, the load of the system, the heterogeneity among hospitals, and, importantly, the method hospitals use to estimate delays. We show that delay estimators that are based on historical averages may cause oscillation in the system and lead to higher average wait times when patients are sensitive to delay. We provide empirical evidence that suggests that such oscillations occur in hospital networks in the United States.
... Another common data-driven approach is to design the predictors in a queuing theoretic context [6,12]. Authors in [6] propose to predict the expected waiting time in a single multi-server queue from the queuing delay history, and compare their approach to common estimators based on the queue length. ...
... Authors in [6] propose to predict the expected waiting time in a single multi-server queue from the queuing delay history, and compare their approach to common estimators based on the queue length. Authors in [12] introduce a framework to predict waiting times in service queues. They consider various predictors, including delay-history-based predictors and snapshot predictors. ...
Conference Paper
Full-text available
With the advent of edge computing, there is increasing interest in wireless latency-critical services. Such applications require the end-to-end delay of the network infrastructure (communication and computation) to be less than a target delay with a certain probability, e.g., 10 −2-10 −5. To deal with this guarantee level, the first step is to predict the transient delay violation probability (DVP) of the packets traversing the network. The guarantee level puts a threshold on the tail of the end-to-end delay distribution; thus, it makes data-driven DVP prediction a challenging task. We propose to use the extreme value mixture model in the mixture density network (MDN) method for this task. We implemented it in a multi-hop queuing-theoretic system to predict the DVP of each packet from the network state variables. This work is a first step toward utilizing the DVP predictions, possibly in the resource allocation scheme or queuing discipline. Numerically, we show that our proposed approach outperforms state-of-the-art Gaussian mixture model-based predictors by orders of magnitude, in particular for scenarios with guarantee levels above 10 −2 .
... Various approaches exist for dealing with incomplete data of processes with nonisolated cases that compete for scarce resources. In call-center processes, thoroughly studied in [12], queueing theory models can be used for load predictions under assumptions about distributions of unobserved parameters, such as customer patience duration [6], while assuming high load snapshot principle predictors show better accuracy [21]. For time predictions in congested systems, the required features are extracted using congestion graphs [20] mined using queuing theory. ...
... Sect. 2) approaches the problem of analyzing the performance of systems with shared resources primarily either from the control-flow perspective [15,17,18,5,9] or the resource/queuing perspective [12,6,21,20] other perspective. In the following, we show how to conceptualize the problem from both perspectives at once using synchronous proclets [11]. ...
Conference Paper
Full-text available
To identify the causes of performance problems or to predict process behavior, it is essential to have correct and complete event data. This is particularly important for distributed systems with shared resources, e.g., one case can block another case competing for the same machine, leading to inter-case dependencies in performance. However, due to a variety of reasons, real-life systems often record only a subset of all events taking place. For example, to reduce costs, the number of sensors is minimized or parts of the system are not connected. To understand and analyze the behavior of processes with shared resources, we aim to reconstruct bounds for timestamps of events that must have happened but were not recorded. We present a novel approach that decomposes system runs into token trajectories of cases and resources that may need to synchronize in the presence of many-to-many relationships. Such relationships occur, for example, in warehouses where packages for N incoming orders are not handled in a single delivery but in M different deliveries. We use linear programming over token trajectories to derive the timestamps of unobserved events in an efficient manner. This helps to complete the event logs and facilitates analysis. We focus on material handling systems like baggage handling systems in airports to illustrate our approach. However, the approach can be applied to other settings where recording is incomplete. The ideas have been implemented in ProM and were evaluated using both synthetic and real-life event logs.
... Process performance has also been approached from the perspective of queuing theory. Senderovich et al. [38,39] propose a method to discover characteristics of "work queues" from event logs at the level of an entire process or of individual activities. In [40], the authors target the analysis of resource-driven processes based on event logs. ...
... In the remaining studies, the authors developed their own applications. [36], [25], [5], [6], [34], [1] Process Duration -- [43] Fragment Duration Activity Start and End Time [7], [26], [25], [19], [5] Activity Duration -- [21], [33], [31], [36], [7], [26] Waiting Duration [32], [40], [38], [39] Delay Duration Internal Quality . ...
Chapter
Process mining has gained traction over the past decade and an impressive body of research has resulted in the introduction of a variety of process mining approaches measuring process performance. Having this set of techniques available, organizations might find it difficult to identify which approach is best suited considering context, performance indicator, and data availability. In light of this challenge, this paper aims at introducing a framework for categorizing and selecting performance analysis approaches based on existing research. We start from a systematic literature review for identifying the existing works discussing how to measure process performance based on information retrieved from event logs. Then, the proposed framework is built starting from the information retrieved from these studies taking into consideration different aspects of performance analysis.
... 6 5 The synthetic logs are publicly available at https://github.com/dafna-s/ Inductive-Context-aware-Process-Discovery/tree/master/Synthetic_log 6 We assume a Poisson arrival process, a common arrival scheme in the literature [27]. A truncated distribution is a conditional distribution that restricts the domain of a given distribution. ...
... Specifically, as a first step, a control-flow model (e.g., a Petri net) is discovered. Then, the model is annotated with decisions [14,31], resources and time [32], queues [27], and general context information [33]. These approaches improve performance analysis and predictive monitoring using the additional context in the enhancement stage, whereas in our work, we focus on improving the quality of the resulting model at the discovery stage. ...
Article
Discovery plays a key role in data-driven analysis of business processes. The vast majority of contemporary discovery algorithms aims at the identification of control-flow constructs. The increase in data richness, however, enables discovery that incorporates the context of process execution beyond the control-flow perspective. A “control-flow first” approach, where context data serves for refinement and annotation, is limited and fails to detect fundamental changes in the control-flow that depend on context data. In this work, we thus propose a novel approach for combining the control-flow and data perspectives under a single roof by extending inductive process discovery. Our approach provides criteria under which context data, handled through unsupervised learning, take priority over control-flow in guiding process discovery. The resulting model is a process tree, in which some operators carry data semantics instead of control-flow semantics. We show that the proposed approach produces trees that are context consistent, deterministic, complete, and can be explainable without a major quality reduction. We evaluate the approach using synthetic and real-world datasets, showing that the resulting models are superior to state-of-the-art discovery methods in terms of measures based on multi-perspective alignments.
... Previous studies focusing on queue length-based delay predictors include Whitt (1999b); Ibrahim & Whitt (2011b) and few other studies discussing delay history-based predictors include Armony et al. (2009) ;Ibrahim & Whitt (2009a, 2011b. For more complex systems, machine learning based delay predictors such as artificial neural networks and decision tree based predictors that are trained on queue log data were proposed in Senderovich et al. (2014Senderovich et al. ( , 2015 and Thiongane et al. (2020). Baldwa et al. (2020) propose a combined simulation and machine learning method for realtime delay prediction for complex queuing systems where adequate queue log data for training a machine learning predictor is not maintained. ...
... The inpatient ward follows a first-come first-served service discipline (i.e., one job type does not have priority over the other). We note here that the majority of the existing literature, as described in Section 2, has focused on developing delay predictors for multi-class multi-server queuing systems with Markovian service times, and where non-Markovian service times have been considered (Nakibly, 2002;Senderovich et al., 2014), job classes with the same type of distribution have been considered (e.g., service times for both job classes follow the Gamma distribution, albeit with different parameters). ...
Preprint
Full-text available
In this work, we demonstrate how real-time delay prediction can be used to determine patient diversion across a healthcare facility network. Specifically, we consider the case of a primary healthcare facility network within the Indian context. We develop a discrete-event simulation of a network of nine primary healthcare centers, each of which provides outpatient care and limited inpatient and childbirth care. We consider diversion for childbirth patients because we observe that under normal operating conditions, a significant proportion of childbirth patients experience substantial wait times before admission. We also use the simulation to determine the conditions under which inpatients may benefit from diversion. We propose diversion mechanisms based on real-time predictions of delays at the facilities from which the patient is being diverted as well as at the facility to which diversion is planned. This involves generating real-time delay predictions not only at the point in time at which the patient arrives at the facility of origin, but also at the point in time in the near future at which the patient may arrive at the facility considered for diversion. We demonstrate the implementation of these diversion mechanisms using actual delays and delay estimates generated by delay predictors based on system state information such as queue length and elapsed service time. As part of this, we develop a novel approximate real-time delay predictor for the queuing system represented by the childbirth and inpatient care processes, and compare its performance to existing delay predictors for this queuing system. With regard to simulation experiments, we first investigate the conditions under which generation of delay predictions in the near future is relevant. We then simulate our proposed diversion mechanism and show that the proportion of patients waiting longer than a threshold wait time decreases with diversion, and that the extent to which operational outcomes become equitably distributed across the PHC networks depends upon the accuracy of the delay predictor. To our knowledge, our study is the first to investigate the use of real-time delay prediction in implementing diversion mechanisms across networks of healthcare facilities.
... However, a clear distinction can be made between reactive approaches [21,33,40], which detect and remove potential problems from the current state of the system, and proactive approaches which predict potential problems and anticipate solutions [15,34]. Among reactive solutions, we distinguish between on user-demand [40] and automatic [21,33] solutions. ...
... Furthermore, they pay no attention to data distribution within the input data streams. Finally, some model-based solutions [15,34] anticipate congestion, thanks to a complete model of the execution support and operator features (processing latencies, pending queues, etc.). Here, the parallelism degrees are adapted to minimize overall latency. ...
... The process mining discipline has provided multitude of tools for analyzing process data from the control-flow perspective. In addition, methods for mining concrete perspectives, such as the queueing perspective were also introduced [12]. Other communities, such as machine learning and data mining, also offer tools to generate models from data. ...
Chapter
The discipline of process mining was inaugurated in the BPM community. It flourished in a world of small(er) data, with roots in the communities of software engineering and databases and applications mainly in organizational and management settings. The introduction of big data, with its volume, velocity, variety, and veracity, and the big strides in data science research and practice pose new challenges to this research field. The paper positions process mining along modern data life cycle, highlighting the challenges and suggesting directions in which data science disciplines (e.g., machine learning) may interact with a renewed process mining agenda.
... For instance, machine learning techniques have been used in [7] to predict flight delays by exploiting available information from sensors in the airport. Combining process mining and queueing-theoretic results, [8] introduces queue-mining techniques for predicting waiting times in service systems. Ang et al. [9] propose a new predictor, called Q-Lasso, which combines the Lasso method from statistical learning and fluid models from queueing theory. ...
... For instance, machine learning techniques have been used in [7] to predict flight delays by exploiting available information from sensors in the airport. Combining process mining and queueing-theoretic results, [8] introduces queue-mining techniques for predicting waiting times in service systems. Ang et al. [9] propose a new predictor, called Q-Lasso, which combines the Lasso method from statistical learning and fluid models from queueing theory. ...
Preprint
Full-text available
Motivated by interest in providing more efficient services in customer service systems, we use statistical learning methods and delay history information to predict the conditional distribution of the customers' waiting times in queueing systems. From the predicted distributions, descriptive statistics of the system such as the mean, variance and percentiles of the waiting times can be obtained, which can be used for delay announcements, SLA conformance and better system management. We model the conditional distributions by mixtures of Gaussians, parameters of which can be estimated using Mixture Density Networks. The evaluations show that exploiting more delay history information can result in much more accurate predictions under realistic time-varying arrival assumptions.
... Additionally, it should be researched whether event processing techniques, which have been already applied in the batch creation phase to adapt clusters, can be also applied in the batch execution phase to allow additional flexibility and to handle exceptions. Recent research efforts focus on process predictions, for instance, on predicting the remaining time [81,94], delays [101], or the process outcome [56]. These types of techniques could support and optimize the batch activation. ...
Thesis
Full-text available
Business process automation improves organizations’ efficiency to perform work. Therefore, a business process is first documented as a process model which then serves as blueprint for a number of process instances representing the execution of specific business cases. In existing business process management systems, process instances run independently from each other. However, in practice, instances are also collected in groups at certain process activities for a combined execution to improve the process performance. Currently, this so-called batch processing is executed manually or supported by external software. Only few research proposals exist to explicitly represent and execute batch processing needs in business process models. These works also lack a comprehensive understanding of requirements. This thesis addresses the described issues by providing a basic concept, called batch activity. It allows an explicit representation of batch processing configurations in process models and provides a corresponding execution semantics, thereby easing automation. The batch activity groups different process instances based on their data context and can synchronize their execution over one or as well multiple process activities. The concept is conceived based on a requirements analysis considering existing literature on batch processing from different domains and industry examples. Further, this thesis provides two extensions: First, a flexible batch configuration concept, based on event processing techniques, is introduced to allow run time adaptations of batch configurations. Second, a concept for collecting and batching activity instances of multiple different process models is given. Thereby, the batch configuration is centrally defined, independently of the process models, which is especially beneficial for organizations with large process model collections. This thesis provides a technical evaluation as well as a validation of the presented concepts. A prototypical implementation in an existing open-source BPMS shows that with a few extensions, batch processing is enabled. Further, it demonstrates that the consolidated view of several work items in one user form can improve work efficiency. The validation, in which the batch activity concept is applied to different use cases in a simulated environment, implies cost-savings for business processes when a suitable batch configuration is used. For the validation, an extensible business process simulator was developed. It enables process designers to study the influence of a batch activity in a process with regards to its performance.
... OCPM assumes that an event may relate to multiple objects corresponding to different case notions. In turn, Queue Mining (QM) relaxes the immediate resource availability assumption [8]. QM addresses situations where multiple cases compete for limited resources, process execution is delayed, and activities are completed only when resources become available. ...
Article
Full-text available
In the Agent-Based Modeling (ABM) paradigm, an organization is a Multi-Agent System (MAS) composed of autonomous agents inducing business processes. Process Mining automates the creation, update, and analysis of explicit business process models based on event data. Process Mining techniques make simplifying assumptions about the processes discovered from data. However, actual business processes are often more complex than those restricted by Process Mining assumptions. Several Process Mining approaches relax these standard assumptions by discovering more realistic process models. These approaches can discover more realistic process models. However, these models are often difficult to visualize and, consequently, to understand. Many MASs induce processes whose behaviors become more complex with each next embraced time step, while the complexities of these MASs remain constant. Thus, the ABM paradigm can cope naturally with the increasing complexity of the discovered process models. This paper proposes Agent System Mining (ASM) and ASM Framework. ASM combines Process Mining and ABM in the Business Process Management (BPM) context to infer MAS models of operational business processes from real-world event data, while ASM Framework maps ASM activities to different phases of the MAS modeling lifecycle. The paper also discusses the benefits of using ASM and outlines challenges associated with the implementation of the ASM Framework.
... Many techniques such as the historical based predictor, Queue-length based predictor, machine learning based predictor, artificial neural network is applied to predict the waiting time [7]- [10]. ...
Article
Full-text available
Queue management is a crucial part of service industry. Business has to deal with the uncertainty of arrival customer and the trade-off between the costs of providing capacity and customer satisfaction. Matching service capacity with the arrival customer is difficult so the system utilization is low in the low periods and customers have to wait for a long time in the peak period. Due to long waiting, some customer abandons the queue before receiving service. Customer abandonment affects business revenue and the system utilization. To relieve the effect of the customer abandonment, this paper aims to propose Artificial Neural Network based waiting time predictor with queue reservation system. The proposed system increases the system utilization and increase customer satisfaction at the same time. Instead of abandonment, customers can reserve their place in queue if the waiting time is too long and they can do other activities while waiting. Moreover, the proposed system provides the accurate estimated waiting time to each customer instead of queue-length. The accurate waiting time enables arrival customers to better decide on the reservation options. 95% of our predicted waiting time is accurate within 5 minutes. Comparing to the system without the queue reservation system, the system utilization and the number of the arrival customer completed service improve 13% and 54% respectively.
... Several approaches have been proposed that focus on timerelated aspects of process prediction. The work in (Pika et al., 2012) describes a technique to predict deadline violations, (Metzger et al., 2015) focuses on predicting delays of activities and (Senderovich et al., 2014) predicts delays of instance executions by applying queue mining techniques. Furthermore, there are approaches that predict the remaining cycle time of running instances (Van der van Dongen et al., 2008). ...
Conference Paper
Full-text available
This study aims to create a queuing system engineering during peak hours. The analysis was done by using queuing theory method with M / M / S calculation. The calculation is to get the average service time, average waiting time, and queue structure. The result of this research is the average service time per person 5 minutes and average waiting per person 13 minutes The standard service time for each customer is two minutes.. The system engineering is done by making 2 queue lines, where the first line for the client queue that carries less than 2 transaction applications with 1 Teller and the second line for queue is more than equal to 3 application transactions with 3 Teller. Queue analysis based on consumer behavior is able to optimize bank services without having to add service servers.
Chapter
In the last decade, stream processing has become a very active research domain motivated by the growing number of stream-based applications. These applications make use of continuous queries, which are processed by a stream processing engine (SPE) to generate timely results given the ephemeral input data. Variations of input data streams, in terms of both volume and distribution of values, have a large impact on computational resource requirements. Dynamic and Automatic Balanced Scaling for Storm (DABS-Storm) is an original solution for handling dynamic adaptation of continuous queries processing according to evolution of input stream properties, while controlling the system stability. Both fluctuations in data volume and distribution of values within data streams are handled by DABS-Storm to adjust the resources usage that best meets processing needs. To achieve this goal, the DABS-Storm holistic approach combines a proactive auto-parallelization algorithm with a latency-aware load balancing strategy.
Experiment Findings
Full-text available
Customer satisfaction rate is one of the essential key performance indicators for the companies. Especially for the companies that offer offline services, in other words the services such as health, pharmacy, car inspection and so on. Moreover, these are the services that have always struggled with queues, delays. Nowadays with extensive development of technology most of these companies widely adopted to Information systems. This incorporation with digitalization has not only switched their paperwork job into digital form, but also has enabled a better management over their data. Obviously in today's competitive world the companies do not wish to fade away, but to continue serving and profiting. The companies are always in search of optimizing their business processes. Apparently, the following questions are more frequently asked ones. 1. How long will it take to complete the certain task? 2. Will we finish the job in a scheduled time? 3. What part of our process is slow or prolongs the process? By using data from those information systems, more precisely event logs we can answer the questions even beforehand. The research fields so called process and queue mining apply the techniques to analyse and discover the reasons for delays, queues and bottlenecks in business processes. In this systematic literature review I will discuss with you data mining techniques to analyse queues on business process execution logs and how do these techniques compare relative to each other.
Article
Predictive process monitoring has recently gained traction in academia and is maturing also in companies. However, with the growing body of research, it might be daunting for companies to navigate in this domain in order to find, provided certain data, what can be predicted and what methods to use. The main objective of this paper is developing a value-driven framework for classifying existing work on predictive process monitoring. This objective is achieved by systematically identifying, categorizing, and analyzing existing approaches for predictive process monitoring. The review is then used to develop a value-driven framework that can support organizations to navigate in the predictive process monitoring field and help them to find value and exploit the opportunities enabled by these analysis techniques.
Article
Business process time prediction aims to predict the time of the running process instance, including the elapsed time and the predicted remaining time. Existing time prediction methods rarely take the cycles of business process into consideration. However, cycles of business process are one of the main factors affecting the accuracy of time prediction. To address this issue, a new transition-driven time prediction for business processes with cycles is proposed. First of all, on the one hand, a Petri net from the event log is mined to obtain the reachability graph, and then the transition division sequence is obtained from the reachability graph of a Petri net. On the other hand, the prefixes are generated by the event log, then, the prefix is feature-encoded and the corresponding business process time is calculated. Second, all prefixes are partitioned according to the activity of the last event of the prefix based on the transition division sequence. Finally, different autoencoders are applied to different transition divisions to reduce dimensionality, and transfer learning is performed by different deep neural networks. Furthermore, by extensive experimental evaluation using the publicly available synthetic event logs and real-life event logs, we show that the proposed method outperforms existing baseline methods.
Article
Full-text available
Predictive business process monitoring methods exploit historical process execution logs to generate predictions about running instances (called cases) of a business process, such as the prediction of the outcome, next activity, or remaining cycle time of a given process case. These insights could be used to support operational managers in taking remedial actions as business processes unfold, e.g., shifting resources from one case onto another to ensure the latter is completed on time. A number of methods to tackle the remaining cycle time prediction problem have been proposed in the literature. However, due to differences in their experimental setup, choice of datasets, evaluation measures, and baselines, the relative merits of each method remain unclear. This article presents a systematic literature review and taxonomy of methods for remaining time prediction in the context of business processes, as well as a cross-benchmark comparison of 16 such methods based on 17 real-life datasets originating from different industry domains.
Article
Business process performance mining offers various analysis views of the performance of a process in a given period of time. However, the analysis of the engineering-to-order (ETO) production system is limited. Stage-based process performance mining has been around to monitor and analyze the process in the predefined stages level. In the ETO production system, it requires more flexibility to analyze the stages. Meanwhile, previous work assumed that the cases should follow the sequence of all predefined stages which is not the case in the ETO production system. This study extends the stage-based process performance analysis by relaxing the definition of the stages, that is referred as a relax-stage-based process performance mining. It emphasizes the flexibility to analyze the process through three different stages: mandatory, necessary, and optional. The concept has been tested in a log data of shipbuilding company. The proof-of-concept of relax-stage-based process performance mining is shown by displaying the time-domain and frequency-domain process performance analysis, including the constructed process model.
Chapter
We employ Mixture Density Networks (MDNs) as a general approach for estimation of customer waiting times in queueing systems based on system states that customers observe upon their arrival. We generate a large amount of data by using discrete-event simulation. Part of the generated dataset is used to train the model, and we utilize the whole dataset for the evaluation of the model. Finally, we illustrate this application in a real-world dataset.
Article
Quantitative business process analysis is a powerful approach for analysing timing properties of a business process, such as the expected waiting time of customers or the utilization rate of resources. Multiple techniques are available for quantitative business process analysis, which all have their own advantages and disadvantages. This paper presents a novel technique, based on queueing models, that combines the advantages of existing techniques, in that it leads to accurate analyses, is computationally inexpensive, and feature complete with respect to its support for basic process modelling constructs. An extensive quantitative evaluation has been performed that compares the presented queueing model to existing queueing models from literature. This evaluation shows that the presented model outperforms existing models with one order of magnitude on accuracy. The resulting queueing model can be used for fast and accurate timing predictions of business process models. These properties are useful in optimization scenarios.
Article
Process Mining is a new kind of Business Analytics and has emerged as a powerful family of Process Science techniques for analysing and improving business processes. Although Process Mining has managerial benefits, such as better decision making, the scientific literature has investigated it mainly from a computer science standpoint and appears to have overlooked various possible applications. We reviewed management-orientated literature on Process Mining and Business Management to assess the state of the art and to pave the way for further research. We built a seven-dimension framework to develop and guide the review. We selected and analysed 145 papers and identified eleven research gaps sorted into four categories. Our findings were formalised in a structured research agenda suggesting twenty-five research questions. We believe that these questions may stimulate the application of Process Mining in promising, albeit little explored, business contexts and in mostly unaddressed managerial areas.
Article
We present an expanded framework for the use of business analytics in projects. To the commonly used descriptive, predictive, and prescriptive analytics, we add comparative analytics, wherein we compare the performance of systems under different interventions. This framework provides a conceptual roadmap for the implementation of business analytics projects. We then demonstrate this framework using recent operations research literature on analytics in healthcare, summarizing papers focusing on one of these aspects. Next, we discuss queue mining as an example of theory and practice illustrative of these aspects. We conclude there is room for further work by operations researchers and management scientists within business analytics projects generally and the healthcare industry more specifically. We argue future work should consider both theory and practice, especially within prescriptive analytics projects, where analysis through the lens of operations research and management science is imperative. We provide some thoughts on the current and future state of operations research and management science in business analytics.
Chapter
Performance mining from event logs is a central task in managing and optimizing business processes. Established analysis techniques work with a single timestamp per event only. However, when available, time interval information enables proper analysis of the duration of individual activities as well as the overall execution runtime. Our novel approach, performance skyline, considers extended events, including start and end timestamps in log files, aiming at the discovery of events that are crucial to the overall duration of real process executions. As first contribution, our method gains a geometrical process representation for traces with interval events by using interval-based methods from sequence pattern mining and performance analysis. Secondly, we introduce the performance skyline, which discovers dominating events considering a given heuristic in this case, event duration. As a third contribution, we propose three techniques for statistical analysis of performance skylines and process trace sets, enabling more accurate process discovery, conformance checking, and process enhancement. Experiments on real event logs demonstrate that our contributions are highly suitable for detecting and analyzing the dominant events of a process.
Article
Full-text available
Predictive analysis in business process monitoring aims at forecasting the future information of a running business process. The prediction is typically made based on the model extracted from historical process execution logs (event logs). In practice, different business domains might require different kinds of predictions. Hence, it is important to have a means for properly specifying the desired prediction tasks, and a mechanism to deal with these various prediction tasks. Although there have been many studies in this area, they mostly focus on a specific prediction task. This work introduces a language for specifying the desired prediction tasks, and this language allows us to express various kinds of prediction tasks. This work also presents a mechanism for automatically creating the corresponding prediction model based on the given specification. Differently from previous studies, instead of focusing on a particular prediction task, we present an approach to deal with various prediction tasks based on the given specification of the desired prediction tasks. We also provide an implementation of the approach which is used to conduct experiments using real-life event logs.
Chapter
Performance analysis from process event logs is a central element of business process management and improvement. Established performance analysis techniques aggregate time-stamped event data to identify bottlenecks or to visualize process performance indicators over time. These aggregation-based techniques are not able to detect and quantify the performance of time-dependent performance patterns such as batches. In this paper, we propose a first technique for mining performance features from the recently introduced performance spectrum. We present an algorithm to detect batches from event logs even in case of batches overlapping with non-batched cases, and we propose several measures to quantify batching performance. Our analysis of public real-life event logs shows that we can detect batches reliably, batching performance differs significantly across processes, across activities within a process, and our technique even allows to detect effective changes to batching policies regarding consistency of processing.
Chapter
To identify the causes of performance problems or to predict process behavior, it is essential to have correct and complete event data. This is particularly important for distributed systems with shared resources, e.g., one case can block another case competing for the same machine, leading to inter-case dependencies in performance. However, due to a variety of reasons, real-life systems often record only a subset of all events taking place. For example, to reduce costs, the number of sensors is minimized or parts of the system are not connected. To understand and analyze the behavior of processes with shared resources, we aim to reconstruct bounds for timestamps of events that must have happened but were not recorded. We present a novel approach that decomposes system runs into token trajectories of cases and resources that may need to synchronize in the presence of many-to-many relationships. Such relationships occur, for example, in warehouses where packages for N incoming orders are not handled in a single delivery but in M different deliveries. We use linear programming over token trajectories to derive the timestamps of unobserved events in an efficient manner. This helps to complete the event logs and facilitates analysis. We focus on material handling systems like baggage handling systems in airports to illustrate our approach. However, the approach can be applied to other settings where recording is incomplete. The ideas have been implemented in ProM and were evaluated using both synthetic and real-life event logs.
Article
Full-text available
This paper investigates the possibility of predicting each customer's waiting time in queue before starting service in a multiserver service system with the first-come first-served service discipline, such as a telephone call center. A predicted waiting-time distribution or an appropriate summary statistic such as the mean or the 90th percentile may be communicated to the customer upon arrival and possibly thereafter in order to improve customer satisfaction. The predicted waiting-time distribution may also be used by the service provider to better manage the service system, e.g., to help decide when to add additional service agents. The possibility of making reliable predictions is enhanced by exploiting information about system state, including the number of customers in the system ahead of the current customer. Additional information beyond the number of customers in the system may be obtained by classifying customers and the service agents to which they are assigned. For nonexponential service times, the elapsed service times of customers in service can often be used to advantage to compute conditional-remaining-service-time distributions. Approximations are proposed to convert the distributions of remaining service times into the distribution of the desired customer waiting time. The analysis reveals the advantage from exploiting additional information.
Conference Paper
Full-text available
Companies realize their services by business processes to stay competitive in a dynamic market environment. In particular, they track the current state of the process to detect undesired deviations, to provide customers with predicted remaining durations, and to improve the ability to schedule resources accordingly. In this setting, we propose an approach to predict remaining process execution time, taking into account passed time since the last observed event. While existing approaches update predictions only upon event arrival and subtract elapsed time from the latest predictions, our method also considers expected events that have not yet occurred, resulting in better prediction quality. Moreover, the prediction approach is based on the Petri net formalism and is able to model concurrency appropriately. We present the algorithm and its implementation in ProM and compare its predictive performance to state-of-the-art approaches in simulated experiments and in an industry case study.
Article
Full-text available
A call center is a service network in which agents provide telephone-based services. Customers who seek these services are delayed in tele-queues. This article summarizes an analysis of a unique record of call center operations. The data comprise a complete operational history of a small banking call center, call by call, over a full year. Taking the perspective of queueing theory, we decompose the service process into three fundamental components: arrivals, customer patience, and service durations. Each component involves different basic mathematical structures and requires a different style of statistical analysis. Some of the key empirical results are sketched, along with descriptions of the varied techniques required. Several statistical techniques are developed for analysis of the basic components. One of these techniques is a test that a point process is a Poisson process. Another involves estimation of the mean function in a nonparametric regression with lognormal errors. A new graphical technique is introduced for nonparametric hazard rate estimation with censored data. Models are developed and implemented for forecasting of Poisson arrival rates. Finally, the article surveys how the characteristics deduced from the statistical analyses form the building blocks for theoretically interesting and practically useful mathematical models for call center operations.
Article
Full-text available
Queueing environment and feedback regarding the likely magnitude of the delay can also influence customer attitudes and ultimately, in many instances, a firm's market share. Even if we focus on the wait itself, the 'outcome' of the queueing experience may vary nonlinearly with the delay, thus reducing the importance of average time in queue, the traditional measure of queueing performance. This speculative paper uses personal experiences, published and unpublished cases, and occasionally 'the literature' to begin to organize our thoughts on the important attributes of queueing.
Conference Paper
Full-text available
Although simulation is typically considered as relevant and highly applicable, in reality the use of simulation is limited. Many orga- nizations have tried to use simulation to analyze their business processes at some stage. However, few are using simulation in a structured and efiective manner. This may be caused by a lack of training and limita- tions of existing tools, but in this paper we will argue that there are also several additional and more fundamental problems. First of all, the focus is mainly on design while managers would also like to use simu- lation for operational decision making (solving the concrete problem at hand rather than some abstract future problem). Second, there is limited support for using existing artifacts such as historical data and work∞ow schemas. Third, the behavior of resources is modeled in a rather naive manner. This paper focuses on the latter problem. It proposes a new way of characterizing resource availability. The ideas are described and analyzed using CPN Tools. Experiments show that it is indeed possi- ble to capture human behavior in business processes in a much better way. By incorporating better resource characterizations in contemporary tools, business process simulation can flnally deliver on its outstanding promise.
Article
Full-text available
Telephone call centers are an integral part of many businesses, and their economic role is significant and growing. They are also fascinating sociotechnical systems in which the behavior of customers and employees is closely intertwined with physical performance measures. In these environments traditional operational models are of great value---and at the same time fundamentally limited---in their ability to characterize system performance. We review the state of research on telephone call centers. We begin with a tutorial on how call centers function and proceed to survey academic research devoted to the management of their operations. We then outline important problems that have not been addressed and identify promising directions for future research.
Book
Full-text available
An exploration of concepts and techniques necessary for diagnosing and correcting the problems that create queues.
Conference Paper
Full-text available
Workflow management systems facilitate the everyday operation of business processes by taking care of the logistic control of work. In contrast to traditional information systems, they attempt to support frequent changes of the workflows at hand. Therefore, the need for analysis methods to verify the correctness of workflows is becoming more prominent. In this chapter we present a method based on Petri nets. This analysis method exploits the structure of the Petri net to find potential errors in the design of the workflow. Moreover, the analysis method allows for the compositional verification of workflows.
Article
Full-text available
Process mining allows for the automated discovery of process models from event logs. These models provide insights and enable various types of model-based analysis. This paper demonstrates that the discovered process models can be extended with information to predict the completion time of running instances. There are many scenarios where it is useful to have reliable time predictions. For example, when a customer phones her insurance company for information about her insurance claim, she can be given an estimate for the remaining processing time. In order to do this, we provide a configurable approach to construct a process model, augment this model with time information learned from earlier instances, and use this to predict e.g. the completion time. To provide meaningful time predictions we use a configurable set of abstractions that allow for a good balance between "overfitting" and "underfitting". The approach has been implemented in ProM and through several experiments using real-life event logs we demonstrate its applicability.
Article
Full-text available
Motivated by interest in making delay announcements to arriving customers who must wait in call centers and related service systems, we study the performance of alternative real-time delay estimators based on recent customer delay experience. The main estimators considered are: (i) the delay of the last customer to enter service (LES), (ii) the delay experienced so far by the customer at the head of the line (HOL), and (iii) the delay experienced by the customer to have arrived most recently among those who have already completed service (RCS). We compare these delay-history estimators to the standard estimator based on the queue length (QL), commonly used in practice, which requires knowledge of the mean interval between successive service completions in addition to the queue length. We characterize performance by the mean squared error (MSE). We do analysis and conduct simulations for the standard GI=M=s multi- server queueing model, emphasizing the case of large s. We obtain analytical results for the conditional distribution of the delay given the observed HOL delay. An approximation to its mean value serves as a reflned estimator. For all three candidate delay estimators, the MSE relative to the square of the mean is asymptotically negligible in the many-server and classical heavy-tra-c limiting regimes.
Article
Full-text available
Contemporary workflow management systems are driven by explicit process models, i.e., a completely specified workflow design is required in order to enact a given workflow process. Creating a workflow design is a complicated time-consuming process and, typically, there are discrepancies between the actual workflow processes and the processes as perceived by the management. Therefore, we have developed techniques for discovering workflow models. The starting point for such techniques is a so-called "workflow log" containing information about the workflow process as it is actually being executed. We present a new algorithm to extract a process model from such a log and represent it in terms of a Petri net. However, we also demonstrate that it is not possible to discover arbitrary workflow processes. We explore a class of workflow processes that can be discovered. We show that the α-algorithm can successfully mine any workflow represented by a so-called SWF-net.
Article
An important interface between stochastic models and actual systems comes in estimating values for model parameters using “real world” data. This interface between models and systems is studied for one of the most elementary stochastic systems, the M/M/1 queue. Estimating arrival rates and service rates results in a notable discrepancy between the state distribution for the model (estimated parameters) and the state distribution for the actual system (known parameters). Also, the expected number of customers in the model is infinite regardless of the (unknown) value of the actual traffic intensity. The truth of this assertion is obvious if one allows estimated traffic intensities to equal or exceed one. However, it is shown that the mean for the model is infinite even if the estimated traffic intensity is restricted to be strictly less than one.
Conference Paper
Discovering predictive models for run-time support is an emerging topic in Process Mining research, which can effectively help optimize business process enactments. However, making accurate estimates is not easy especially when considering fine-grain performance measures (e.g., processing times) on a complex and flexible business process, where performance patterns change over time, depending on both case properties and context factors (e.g., seasonality, workload). We try to face such a situation by using an ad-hoc predictive clustering approach, where different context-related execution scenarios are discovered and modeled accurately via distinct state-aware performance predictors. A readable predictive model is obtained eventually, which can make performance forecasts for any new running process case, by using the predictor of the cluster it is estimated to belong to. The approach was implemented in a system prototype, and validated on a real-life context. Test results confirmed the scalability of the approach, and its efficacy in predicting processing times and associated SLA violations.
Article
We present optimal mean square predictors for queue lengths and delays in the stationary GI/M/m queue, based on a queue length measurement. The development specifies interrelationships among these predictors and numerical examples demonstrate basic properties of the predictors.
Article
Both empirical evidence and logic suggest that there is a strong negative correlation between waiting time and a customer's evaluation of the quality of a service. The evaluation of service, in turn, is related strongly to customers' loyalty and other important outcomes. A conceptual model, based on field theory, is developed. The model integrates key variables derived from recent studies of consumer waiting behavior. Also incorporated are relevant constructs from the extant services literature, including the roles of the disconfirmation of consumers' wait time expectations, prior service encounters, and the quality of the customer's encounter with the contact employee. Finally, data from actual consumers in a natural queueing context are used to test the theoretical framework. Analysis of the data, with the use of multiple regression, demonstrates a strong pattern of support for the field-theory-based hypotheses, confirming the important role of queue wait management in customer evaluations of service quality. © 1998 John Wiley & Sons, Inc.
Book
The more challenging case of transient analysis of Markov chains is investigated in Chapter 5. The chapter introduces symbolic solutions in simple cases such as small or very regular state spaces. In general, numerical techniques are more suitable and are therefore covered in detail. Uniformization and some variants thereof are introduced as the method of choice for transient analysis in most cases. Particular emphasis is given to stiffness tolerant uniformization that can be of practical relevance in many modeling scenarios where relatively rare and fast events occur concurrently. As an alternative a method for aggregation/disaggregation of stiff Markov chains is introduced for a computation of approximate transient state probabilities. The method is based on a distinction of fast recurrent and fast transient sets of states that can be aggregated with relatively small error. All steps are illustrated by a detailed example model of server breakdown and repair. In addition to numerical results an error analysis is provided as well.
Book
Cambridge Core - Knowledge Management, Databases and Data Mining - Introduction to Information Retrieval - by Christopher D. Manning
Article
The stochastic processes which occur in the theory of queues are in general not Markovian and special methods are required for their analysis. In many cases the problem can be greatly simplified by restricting attention to an imbedded Markov chain. In this paper some recent work on single-server queues is first reviewed from this standpoint, and the method is then applied to the analysis of the following many-server queuing-system: Input: the inter-arrival times are independently and identically distributed in an arbitrary manner. Queue-discipline: "first come, first served." Service-mechanism: a general number, $s$, of servers; negative-exponential service-times. If $Q$ is the number of people waiting at an instant just preceding the arrival of a new customer, and if $w$ is the waiting time of an arbitrary customer, then it will be shown that the equilibrium distribution of $Q$ is a geometric series mixed with a concentration at $Q = 0$ and that the equilibrium distribution of $w$ is a negative-exponential distribution mixed with a concentration at $w = 0$. (In the particular case of a single server this property of the waiting-time distribution was first discovered by W. L. Smith.) The paper concludes with detailed formulae and numerical results for the following particular cases: Numbers of servers: s = 1, 2 and 3. Types of input: (i) Poissonian and (ii) regular.
Article
Consider a feedforward network of single-server stations populated by multiple job types. Each job requires the completion of a number of tasks whose order of execution is determined by a set of deterministic precedence constraints. The precedence requirements allow some tasks to be done in parallel (in which case tasks would "fork") and require that others be processed sequentially (where tasks may "join"). Jobs of a given type share the same precedence constraints, interarrival time distributions and service time distributions, but these characteristics may vary across different job types. We show that the heavy traffic limit of certain processes associated with heterogeneous fork-join networks can be expressed as a semimartingale reflected Brownian motion with polyhedral state space. The polyhedral region typically has many more faces than its dimension, and the description of the state space becomes quite complicated in this setting. One can interpret the proliferation of additional faces in heterogeneous fork-join networks as (i) articulations of the fork and join constraints and (ii) consequences of the disordering effects that occur when jobs fork and join in their sojourns through the network.
Predicting waiting times in telephone service systems
  • E Nakibly
Nakibly, E.: Predicting waiting times in telephone service systems. Master's thesis, Technion-Israel Institute of Technology (2002)