Predictive Process Monitoring: a comparison survey between different type of event logs

Simona Fioretto, Elio Masciari · 2025

Read Predictive Process Monitoring: a comparison survey between different type of event logs by Simona Fioretto, Elio Masciari (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-16933.txt.

Introduction

Simona Fioretto and Elio Masciari survey methods that predict how business processes will unfold, in a 2025 review paper. Firms use these methods to see delays and faults before they happen. Many methods exist, and they differ in the records they read and the tasks they handle.

The review pays close attention to a newer record type that tracks several linked objects at once, rather than one case at a time. The authors sort the methods by the task they solve and the tools they use. Their map helps readers pick a method that fits their data. It also shows where the newer record type still lacks tested methods.

Catalog record

CREATORSimona Fioretto, Elio Masciari
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-16933

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Predictive Process Monitoring: a comparison survey between different type of event logs Simona Fioretto, Elio Masciari Abstract The application of Predictive Process Monitoring (PPM) techniques is becoming increasingly widespread due to their capacity to provide organizations with accurate predictions regarding the future behavior of business processes, thereby facilitating more informed decision-making. A plethora of solutions have been proposed in the literature employing these techniques, yet they differ from one another due to a number of factors. However, in light of the growing recognition of the value of object-centric event logs, including in the context of PPM, this survey focuses on the differences among PPM techniques employed with different event logs, namely traditional event logs and object-centric event logs. In addition, the reviewed methods are classified according to the prediction task they address and the specific methodologies they employ. 1 Introduction In the context of the ongoing digital transformation, the growing volume of operational data stored within information systems is emerging as a significant business opportunity, offering potential for the development of advanced applications such as information retrieval. Event records of processes execution widely known as Event Logs [ 7 ] can reveal important insights about business processes, thus enabling the development and application of Process Mining (PM) research area [ 65 ] . Moreover, the rising adoption of Statistical Models, Machine Learning techniques, and AI in business solutions [ 6 ] contributes to the advancement of PM area. The process of managing business processes is defined by the field of Business Process Management (BPM) [ 29 ] . The management of a business process is a cyclical process, beginning with the identification of the process itself, followed by the discovery and analysis of the process, then the redesign and implementation of the process, and finally the monitoring of the process. Process monitoring enables real-time extraction of information on the actual execution of the process, facilitating the acquisition of knowledge on process measures and performance, and identifying potential problems. The monitoring phase can be conducted in either an offline or online mode, thus enabling a transition from the application of a-posteriori techniques based on historical data, to the utilization of runtime techniques while the process is ongoing. This paper focuses on Predictive Process Monitoring (PPM) approaches [ 28 ] , which support online process monitoring within the context of Process Mining. PPM offers insight into the future characteristics of ongoing cases by training a prediction model on the event log data of completed cases. In this survey, we focus on the identification of PPM methodologies based on the specific input event logs that are being processed. In this context, we differentiate between two primary categories: the first encompasses traditional event logs, and the second, object-centric event logs, which, to the best of our knowledge, has not been previously considered. This study elucidates the methodologies deployed to integrate object-centric event log data into classical event log PPM methodologies. Through a comprehensive analysis, it demonstrates the significance of utilising this information to enhance prediction accuracy. Furthermore, it provides insights into the potential of adapting existing methodologies for novel applications. The paper is structured as follow: first, an overview of the meaning of predictive process monitoring will be provided, and the difference between an object-centric event log and a classical event log will be highlighted in Section 2 . Then it will be provided a classification of the existing surveys in Section 3 . In Section 4 the methodology employed for the retrieval of the studies is analysed, as is the classification system used to categorise them in Section 5 . 2 Overview In this section we provide an overview of the concepts of Event Log and Predictive Process Monitoring. 2.1 Predictive Process Monitoring Predictive Process Monitoring (PPM) is a specific research area of Process Mining (PM) that employs historical complete process execution traces to make predictions about the future evolution of features of an uncompleted execution trace of an ongoing process [ 39 , 26 ] . By leveraging PPM it is possible to predict the performance indicators of a running process, such as the outcome of a process execution, its completion time, or the sequence of its future activities to cite a few. Event Log traces, i.e. completed process executions are used to train a Predictive Model, which is in general a Machine Learning model trained on historical data X to assign a label on new data Y in the context of prediction. The main objective of the model is to generalize the pattern identified in the training data, providing a mapping X → Y. [ 67 ] .More in detail, given the previous concepts, it is then possible to define the general PPM framework. This framework starts from event logs data containing information on the history of the process (i.e., historical complete process executions) which are given as input for the specific prediction algorithm to be implemented. The algorithm is trained to discover patterns in data and then generalize them. Finally, during the execution of the process, the previously trained model takes as input the event stream or prefix logs, which are live data representing the ongoing trace of a runtime process, to perform the prediction task, thus supporting decisions about the ongoing and future execution. 2.2 Event Log Process Mining approaches rely on the availability of Event Logs in Information Systems, data in which the execution of processes is recorded [ 66 , 2 ] . The Event Log utilized in the context of Process Mining has a specific structure. In order to be utilized, it needs to have three mandatory fields about the process execution. These are respectively • the specific process instance, called case ; • the actual action being recorded, called activity ; • the timestamp of the activity being recorded; • optional attributes that could be recorded during the execution of the process (i.e., used resources, involved people, incurred costs, etc.) The aforementioned specification includes the existence of a unique case notion, with each event linked to exactly one case identifier [ 2 ] . However, in real-world organizational scenarios, this assumption may be insufficient to properly model the data [ 2 , 4 , 3 ] . To address this Object-Centric Event Logs are introduced, to ease the previous constraints. The importance of enhancing data quality and reliability in process mining applications might pursue integrity checking practices [ 17 , 18 , 41 , 42 , 23 , 24 , 22 , 43 ] , top- k selections [ 45 , 19 , 20 , 44 , 15 ] to retain the most promising outcomes, or even leverage collective input from multiple stakeholders to accurately map out processes as in crowd sourcing [ 38 , 32 , 9 ] . We also observe that accessing process mining logs poses challenges similar to those that were addressed in the field of query processing under access limitations [ 12 , 11 ] . Object-Centric Event Logs are used to model and preserve information from real-world scenarios, where data often take the form of graph structures characterized by the interaction and synchronization of multiple objects. This means that a single business process cannot be tied to just one

Read the complete text