Abstract
Part II examined ways to increase process realism through conformance checking. As a result, the focus has so far been mainly on process models, and how to measure their quality. In this part, we will shift focus towards the process data itself. Indeed, as a process is more than control-flow alone, progressing towards a realistic understanding of a process requires more instruments than process models and quality measures alone.
Reproducibilty is actually all aboutbeing as lazy as possible.
Hadley Wickham
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
- 2.
Both R and Python are two widely-used ecosystems for data science, each with their own distinguishing strengths. While Python is increasingly well-known for it’s machine learning techniques, R offers an extremely large repertoire of statistical techniques, and is powerful in the field of data visualisation. In recent years, it is believed by many scholars and practitioners that neither R nor Python will emerge as a predominant language. Instead, the two are increasingly being connected with each other. IDEs such as RStudio are now able to execute both languages, among others such as SQL. Rmarkdown documents even allow one to combine both R and Python code chunks within the same document.
- 3.
It should be noted that processanimateR is an extension which was contributed by Felix Mannhardt [102].
- 4.
Note that the \(\%>\%\)-symbol is called the piping symbol, and is used to pass-through an object as first argument to the following function [16].
- 5.
One might prefer to keep the original, low-level activity instances as events of the newly created activity instance. However, this is not done by the default aggregation, since these low-level events will not fit the transactional lifecycle (Fig. 7.3), as is usually the case when an activity instance contains more than one event—e.g. as the start and complete events of several instances in the example in Table 2.1. In cases where retaining the original activity instances as underlying events is favourable, it is advised to perform a custom aggregation using the general event data processing tools discussed in Sect. 7.5.2.4.
- 6.
For an overview of available metrics, see further in Sect. 7.5.3.
- 7.
Note that there is a fifth filter method for the time period filter, i.e. trim, but this is actually an event filter and will thus be discussed in the next section.
- 8.
It’s important to note that throughout bupaR the term trace is used to refer to a unique sequence of activity labels—also known as a process variant. It is different from the term case, which refers to a single, unique execution of the process.
- 9.
Note that the duration of an activity instance with only one event is equal to zero.
- 10.
- 11.
- 12.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Janssenswillen, G. (2021). Reproducible Process Analytics. In: Unearthing the Real Process Behind the Event Data. Lecture Notes in Business Information Processing, vol 412. Springer, Cham. https://doi.org/10.1007/978-3-030-70733-0_7
Download citation
DOI: https://doi.org/10.1007/978-3-030-70733-0_7
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-70732-3
Online ISBN: 978-3-030-70733-0
eBook Packages: Computer ScienceComputer Science (R0)