Essentially, the following sets of algorithms are provided for the purpose of the work done in this book by considering the different phases that constitutes practical implementation of the SPMaAF framework.
Algorithm 1
The work describes in this section the proposed Algorithm 1 and how it makes use of the method to perform the process mining and model discovery (Phase 1). Perhaps, the algorithm (Algorithm 1) is developed to show how to effectively discover useful process models from the readily available events (data) logs. In principle, the process proves useful towards generation and mapping of the individual traces that makes up each of the process executions. For example, we illustrate in Chapters 5 and 7 how the proposed algorithm 1 is implemented using process mining tools such as Disco that is based on Fuzzy Miner framework (Rozinat & Gunther, 2012) to generate and map the process models from the readily available event logs. In addition to the process models discovery, the process is also carried out for conformance checking and analysis of the individual cases (i.e. classified traces) and visualization of the several sequence(s) of activities executions.
Practically, the following Algorithm 1 describes how the work discovers and generates the process models and individual traces from any given events data log as follows:
1: For all Recorded Events Data Log, L
2: Input:PM – Process mining tool used to extract model, M L – Input Data for process Mapping/Visualization
e – Classifier for the event logs, L and traces, T
3: Assign: case_id(e) i.e. the Case associated to event, e within theevents log, L act_name(e) i.e. Activities associated to event, e within L other_attributes e.g. Event ID, Timestamp, Resources, Roles etc. related to event, e within L
4: Output: Process maps (fuzzy model), M & individuals traces, T classifications for the events log, LModel or TraceFitness, TF discovery through semantic fuzzy mining
5: Procedure: Discover Fuzzy Models, M from L for cross-validation to determine how well M reflects the performed activities in reality, i.e TraceFitness, TF and for further analysis
6: Begin
7: For all Event Data Log L
8: Extract Process Maps, M, &Traces, T ← from Event Log L
9: while no more process element is left do
10: Analyze Fuzzy Model, M and Traces, T to determine tracesFitness, TF
11: IfT ← Null then
12: obtain the occurring act_name(e) sequence sets from Log L
13: ElseIfT ← 1 then
14: cross-validate resulting Trace, T from L with Fuzzy Model, M
15: If trace, T exist then
16: For each event Classifier, e output← return as True_Positive, TP i.e fits the Fuzzy Model, M
17: ElseIf trace, T does not exist then
Return event Classifier, eoutputas True_Negative, TN i.e does not fit the Fuzzy Model, M
18: Return: Classification Results of the Semantic Fuzzy Mining approach and Process Maps
19: End If statements
20: End while
21: End For
Ultimately, from the proposed Algorithm 1, and as earlier explained in the description of the phase 1 of the SPMaAF framework in chapter 3, we recognize that:
- ▪
A typical process model, M consists of Traces, T (i.e. Cases)
- ▪
A Trace (Case), T, consist of events, e, such that each event relates to precisely one case.
- ▪
Events, e, within a Trace are ordered, most often in a sequential order
- ▪
Events for any process mining task must have atleast a Case identification Id (Case_id) and Activity Name (Act_name) attributes to allow for the process model discovery to follow.
- ▪
Other additional information may be required for ample implementation of process mining e.g. Event ID, Timestamp, Resources, Cost, Roles, and Places etc. (Van der Aalst, 2016)