01 Process Mining From Theory to Execution

2023-03-24

위 벳지는 수강을 완료하고 받은 뱃지입니다.

Process Mining: From Theory to Execution

Discover the science behind the technology. Perform real-life applications.

01. Welcome from Professor Wil van der Aalst

Welcome From Professor Wil van der Aalst

Welcome from your hosts, Professor Wil van der Aalst and Rosie from Celonis.

Welcome from Professor Wil van der Aalst

Hello and Welcome to “Process Mining: From Theory to Execution”!

We are happy to welcome you to this joint class by RWTH Aachen and Celonis together with the Godfather of Process Mining, Prof. Wil van der Aalst. This course is designed for everyone who wants to gain an advanced theoretical understanding of Process Mining and how the science behind it comes into application in state-of-the-art technology.

Below you can find all the relevant information that will help you navigate through this class and earn your learner’s badge at the end! After this you can directly go back to the main navigation page and get started with Lecture 1!

Topics and Structure

The “Process Mining: From Theory to Execution” class is brought to you by RWTH Aachen and Celonis and led by the Godfather of Process Mining Prof.dr.ir. Wil van der Aalst.

The goal of this class is to get you familiar with the solid theoretical and academic foundation of Process Mining. With this theoretical knowledge, we will guide you further to the practical application of Process Mining with real business cases, data and insider tips from Celonis. After finishing all the lectures in this class you will understand the most important concepts behind Process Mining and the respective software functionalities building on them and what they can be applied for in a business context.

Let’s meet your instructors Prof.dr.ir. Wil van der Aalst (you can call him Wil ;)) from RWTH Aachen and Rosie from Celonis!

Course Structure

The “Process Mining: From Theory to Execution” class consists of 7 lectures. Each lecture gives you a holistic overview of the topic consisting of three different learning items:

1. Lectures by Wil - here you get familiar with the theoretical side of Process Mining

2. Hands-on Illustration by Celonis - after learning the theory, you are welcome to apply your fresh knowledge in practice by following the practical videos by Celonis

3. Knowledge-Check - each lecture ends with a small quiz to test your knowledge

What’s Next?

After successfully completing the class, you will receive your learner’s badge for this class! We work together with Credly to issue internationally recognized learner’s certifications.

In order to get your badge, you need to complete the Process Mining journey by watching all the lectures in this class and submitting the knowledge-checks. Most importantly, the class counts as successfully completed if you score 50% and higher in the knowledge-checks.

This badge is issued in the name of RWTH Aachen and Celonis. You may add this digital badge to your CV or your LinkedIn Profile.

Good luck!

Meet your Professor

The Godfather of Process Mining

Prof.dr.ir. Wil van der Aalst is a full professor at RWTH Aachen University leading the Process and Data Science (PADS) group. He is also part-time affiliated with the Fraunhofer-Institut für Angewandte Informationstechnik (FIT) where he leads FIT’s Process Mining group and the Technische Universiteit Eindhoven (TU/e). His research interests include Process Mining, Petri Nets, business process management, workflow management, process modeling, and process analysis.

As an academic, Wil is passionate about sharing his knowledge of Process Mining and teaching others. It was one of the motivations behind this class - to provide a learning opportunity bridging the gap between theory and software application and to give you a solid understanding of Process Mining directly by the person who invented it!

We call Wil van der Aalst the Godfather of Process Mining - and there is a good reason for that! Wil has been researching Process Mining for many years in the academic area. Thanks to his work Process Mining found its way into business and built a solid foundation for companies such as Celonis, where Wil holds a position as the Chief Scientific Advisor.

Meet Celonis

Who or what is Celonis?

Celonis is the marketleader for Process Mining and Execution Management Technology facilitating smooth operations for over 2000 enterprises worldwide including 50% of Fortune 500 companies. Celonis has won multiple awards for its outstanding technology including The EY Entrepreneur of the Year Award, the Bain Game Changers Award 2019, 2016 and the German President’s Award 2019. By now Celonis is the most valuable start up company in Germany and New York being valued at 11bn US Dollars having earned the rare Decacorn status.

Celonis was founded in 2011 as a student project of three students from the Technical University of Munich. The three founders Bastian Nominacher, Alexander Rinke & Martin Klenk worked on a student consulting project helping the Bavarian Broadcasting service to improve their IT Service Desk answering times when they stumbled across Process Mining as an academic idea and discipline pioneered by Prof. Wil van der Aalst. Not only Process Mining but also Celonis has an inherently academic DNA. Since then Celonis has turned into one of the fastest-growing companies within Europe, now has two headquarters in Munich and New York and has even turned into a Decacorn with over 2000 employees and an evaluation of 11bn US Dollars recently.

If you want to learn more about Celonis as a Company Check out this article in Forbes telling the story behind the once-in-a-generation company (Forbes, June 2021: https://www.forbes.com/sites/alexkonrad/2021/06/02/celonis-process-mining-raises-at-11-billion-valuation/?sh=a35a6232acd3(opens in a new tab))

The Celonis Execution Management System

The Celonis Software is also known as the Celonis Execution Management System (short EMS). The Celonis EMS has Process Mining technology at its core. Just like the field of Process Mining the Celonis EMS has developed from an historical analytical tool towards an action-oriented operational backbone in enterprises. The Celonis EMS now also integrates prediction capabilities as well as action flows and automation capabilities to prevent bottlenecks from happening in real time.

The Celonis Software includes the chance to upload data from scratch or build direct connectors to source systems (Event Collection), a Celonis Studio to build analyses and integrate Action Flows and a Machine Learning Workbench to run predictions and even import e.g. Python or R packages.

Throughout the course of the lecture we will get to know all of these features in more detail!

Resources

Useful Material

In this section you can find all the useful material that will help you successfully complete this course. Mainly you will need the following:

1. Lecture Slides
2. Celonis EMS Training License
3. Demo Data Sets
4. Getting your data into the data model

In addition you might also be interested in

1. Wil’s Book “Process Mining: Data Science in Action”
2. Advanced Celonis Software Trainings.

LECTURE SLIDES

Lecture1_Introduction_to_PM.pdf

Lecture2_Process_Discovery1.pdf

Lecture3_Process_Discovery2.pdf

Lecture4_Conformance_Checking.pdf

Lecture5_Getting_The_Data.pdf

Lecture6_Comparative_and_Predictive_PM.pdf

CELONIS EMS TRAINING LICENSE & ACCESS INSTRUCTIONS

Throughout this class you’ll have the opportunity to explore Process Mining hands-on in the Celonis software and solve some of the questions from the knowledge checks with it. For this, you need your own EMS Trainining License. Follow the steps in this guide to get access to your free Training Environment:

Accessing your Celonis Training Team.pdf

DEMO DATA SETS

As soon as your Training Environment is all set, you are ready to conduct your own Process Mining analyses! For this, we’ve created a few datasets that you will work with during the lectures:

concurrency-matters.xlsx

concurrency_matters_dev.xlsx

concurrency-matters-skips.xlsx

event_log-12666-orders-simplified.csv

P2P-Activities.csv

P2P-Cases.csv

LEARN HOW TO GET YOUR DATA INTO THE DATA MODEL

FURTHER ACADEMIC READING - THE PROCESS MINING BIBLE

Get the “Process Mining bible” for a deeper dive into academic Process Mining theory.

“PROCESS MINING: DATA SCIENCE IN ACTION” BOOK BY WIL VAN DER AALST

Wil van der Aalst is the author of the so-called Process Mining bible: “Process Mining: Data Science in Action”.

As a learner on this course you are eligible to a 50% discount on the e-book version of this book. You can find the book under the following link: https://bit.ly/2Sgd9UM(opens in a new tab). Use the following code to get the discount: zDddBawnhsmuAUpD.

If you want to get a hard copy of the book, follow this link: https://bit.ly/2NptaqQ(opens in a new tab). For a 20% discount use the following code: vyErXDACaXwmq4Sb.

ADVANCED CELONIS SOFTWARE TRAINING

Basic Coding with PQL

Do you want to learn more about the coding language used in Celonis? PQL (Process Query Language) forms the foundation of analysis building in Celonis.

GO TO COURSE

Action Flows Training

Process Mining and Automation just fit together. Learn about how to create automatic action flows with Celonis.

GO TO COURSE

02. Overview of the Field and Basic Concepts

Introduction: Process Mining From Theory to Execution

Introduction to Process Mining and its multiple facets.

Welcome to your first lecture of “Process Mining: From Theory to Execution”. In this first Lecture we will start with an Introduction to the main themes currently relevant in Process Mining. The aim of this first lecture unit is to give you an overview of what Process Mining entails and where it currently stands as well as to introduce you to some basic terminologies and concepts that are important for the field.

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Start your Process Mining journey now!

Lecture 1: Introduction to Themes

“Process Mining is a complex and continually expanding field of study. From its conception as a research topic in the late 90’s it has grown into a globally applied science touching the likes of more than half of fortune 500 companies. “

Introduction to Themes

Wil van der Aalst pioneered the field of Process Mining in the 1990s. He was the first to investigate the missing link between process and data science. In the first part of this course he will give you an overview of the field and introduce you to important concepts and terminologies.

In this first lecture, Wil will introduce you to the fundamentals of Process Mining and a top-down overview of its structure. You will learn about different tasks in Process Mining and what is needed to accomplish them. Furthermore, after watching this lecture, you will get familiar with the most important attributes of data required for Process Mining: Case, Activity, and Timestamp. By using these types of data, you will not only get valuable insights into the as-is processes, but also will be able to make predictions to optimize processes in the future.

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

By now, you should have learned the definition, history, and importance of Process Mining for different businesses. You should also be familiar with the overview and the different focus areas that the field holds: From Process Discovery to action-oriented Process Mining. Now, let’s see how you can apply Process Mining in practice together with Celonis - for this, join the hands-on Lecture by Celonis.

Lecture 1: Celonis Hands-On

“There are a few important ingredients that provide function and methodology in the world of Process Mining”.

Introduction

Welcome to your first applied lecture within “Process Mining from Theory to Execution”. Wil has already given you an overview of what Process Mining is and how it helps to make vast amounts of process-related data visible and interpretable.

The purpose of these hands-on exercises and illustrations in Celonis is to show you how academic concepts come into practice in real-life scenarios. There is hardly any other field where academia and industry inspired each other as much as in Process Mining. Techniques for process discovery, conformance checking, and performance analysis emerged from research and were made accessible for a large audience through the Celonis software. In return, large-scale applications of these techniques triggered new research questions. The purpose of this course is to bridge the gap between the two worlds and to show academic concepts in their application in real life. For today we will focus on illustrating some very basic terms and concepts that are important for any Process Mining Analysis. We introduce basic concepts such as event, case, activity, timestamp and event log.

Data can be enormous. Just imagine going back in time the past year to manually collect data about each step you took, when you took it, and where. And what you did next after each step. How would you get it done, and how long would it take you? Luckily, Process Mining can analyze event log data to do just that. There are some fixed terminologies for the analysis of an event log that we want to illustrate today.

Demo Data for this Lecture

You might be curious about what you have learned in this practical lecture. The illustrations in this lecture have all been constructed around the Purchase-to-Pay process in a random organisation. Purchase-to-Pay processes are among the most frequently studied processes in organisations since they are error-prone due to huge volumes of transactions, many departments involved and tight deadlines.

You are encoured to study the demo data yourself in the Celonis Demo below!

You are encoured to study the demo data yourself in Celonis. View the example from the Lecture video here!

DEMO NOW!

In this first chapter you will have gotten a first feeling of what Process Mining and how big its momentum is at the moment. Both in research and industry Process Mining is a blooming field with lots of opportunities.

After this First Lecture you should now be familiar with basic concepts and terminologies such as Process, Event Data and Process Model as well as the concepts of activities, events and cases.

Check your knowledge in the exercise section and see if you have understood both the theory and hands-on aspects of this lecture!

Lecture 1: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in a real world setting and transfer your knowledge to other use cases. You must complete all Knowledge Check sections in order to obtain your certificate!

Task 1: Analyse Variants of Traces.

Have a look at the event log snippet below. Here you can see various purchase orders requested and created by Celonis employees recently. Answer the following questions:

Task 1.1. How many different variants can you identify in the event log?

• 3
• 4 (o)
• 5

Task 1.2. How does the most frequent variant in the event log look like? (Hint: It occurs in two cases). Bring the activities into the right chronological order based on the timestamps.

[1] Create Purchase Order

[2] Print and send Purchase Order

[3] Goods receipt

[4] Book Invoice

Task 2: Analyse the Purchase To Pay Process.

Now let’s look at larger event log that has already been connected to the Process Mining engine of the Celonis Execution Management System. Go back to the Purchase to Pay (P2P) analysis from the Celonis Video and answer the questions below. You can access the P2P example here: https://content.training.celonis.cloud/process-mining/public/6cdb3291-b77d-4a0a-bb5a-13a20c028feb/#/frontend/documents/6cdb3291-b77d-4a0a-bb5a-13a20c028feb/view/sheets/333dbdf2-94e5-41d2-95dc-0e26b97b6738

How frequently does each of the listed process variants occur in the Purchase to Pay process?

How many activities does the happy path (most frequent process variant) consist of?

• 4

• 6

• 9

Tick the right statements.

• An activity only ever shows up in one single processes.

• Process activities are actions that initiate or terminate a process, or take place during it.

• The Procurement process covers activities of requesting, purchasing, receiving, and paying for goods and services.

• Typically, undesired Procurement activities are: Change Price, Create Purchase Requisition, Book Invoice.

Task 3: Distinguish 3 Basic Concepts.

Match these three basic concepts and terminologies in Process Mining to their right definition.

Match the right definition for…

[3] Event Log

[2] Process Model

[1] Process

1 Series of linked activities, taken in order to achieve a particular goal

2 Describes a process and is based on the corresponding event data

3 List of events related to the process

03. Process Discovery and Directly-Follows Graphs

Process Discovery: Directly-Follows Graphs

Understand process discovery and how to get from data to the graph.

WHAT’S INCLUDED

Self-paced learning. Take courses at your own speed.

Learn when it suits you with round-the-clock access.

Get enabled fast with laser-focused, goal-based training.

Personal Training Environment Required

This course contains hands-on exercises that require a personal EMS Training Environment, also called an “EMS Team”.

Not sure if you have one? Click below and we’ll either create one for you or show you how to access your existing Team if you have one. Please make sure to disable adblockers on this page if the button is not working for you.

Thank you for your request! Your personal environment should be created within 10-15 minutes. Please check your inbox for an activation link from the sender “no-reply@celonis.cloud”. Note that in very high traffic times the creation can take up to 1.5 hour.

Once activated, you can always access your team by logging in to Celonis ID - This is a central login that shows you all of the EMS Teams you have access to. You can also simply bookmark the link your personal training team since you will be asked to repeatedly use it in a multitude of Celonis courses. Feel free to continue with your training now.

Lecture 2: Process Discovery (1/2) - Learning Directly-Follows Graphs

Welcome to the first part of the Process Discovery chapter and the second Lecture of “Process Mining: From Theory to Execution”. In this lecture unit we will take a step further to discover basic process models. We will focus on Directly-Follows Graphs. The aim of this lecture unit is to give you an overview of the basic process models, how to use them and when to apply different filtering techniques to get a clearer view of your model.

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Let’s take our second step in the Process Mining Journey now!

Lecture 2: Process Discovery (1/2) - Learning Directly-Follows Graphs

“Directly-Follows Graphs are an extremely scalable way to showing process flows. If you have hundreds of millions of events these can be analyzed without a problem”

Let’s look at Directly-Follows Graphs!

The second lecture is dedicated to a fundamental concept for Process Discovery- Directly-Follows Graphs (DFG). Representing the processes in a very straightforward way, DFGs are a powerful tool to get insights from raw event data. After watching this lecture you will be able to define what a DFG is, understand its basic features, and apply different filtering techniques.

More specifically, you will learn:

how activities and connections complement each other to give you an overview of a process

• what is the difference between a frequency view and a time view
• how DFGs can help you spot bottlenecks in your processes
• how to differentiate between and combine three filtering types to focus on what is important for you.

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

In this lecture you have learnt that DFG is a very powerful tool to show frequencies and spot bottlenecks in processes. In fact DFGs are one of the most frequently used process representations also in industry. They are very intuitive and also allow for a story-telling element as we see the different process variants unfold as the process really happened based on variant frequencies.

Now, it is time to see how DFGs are realized in the Celonis software! Let’s move on to the next Practical Lecture by Celonis to explore the power of process models with our example dataset and build your own DFG model!

Lecture 2: Celonis Hands-On

“By looking at all process variants we can see that processes quickly get complex. Filtering methods help to get a better overview of the relevant process variants.”

Directly-Follows Graphs in Celonis

The big challenge behind Process Mining is to go from an event log to a process graph. There are several methods to do this and we will learn all about them in the course of the lecture.

One of the most straightforward ways to create a process graph, is to use Directly-Follows Graphs. In Celonis Directly-Follows Graphs are shown in the form of either the Celonis Variant or the Celonis Process Explorer. The Variant and Process Explorer allow you to visualize the as-is process flow. The process flow is based on the sequence of the activities and the activities themselves that have taken place.

We will look at the Variant and Process Explorer in more detail now. This will reveal how to get from the process data stored in the event log to a visual representation in a process graph. In the next step we will also discuss different types of filtering, namely activity and arc-based filtering. By learning about these filters we will understand how they affect our perspective on a process.

Before jumping into the practical part by Celonis, make sure that your Training environment is all set up:

Accessing your Celonis Training Team.pdf

In this lecture, we will work with the following data set. Download it before watching the video:

event_log-12666-orders-simplified.csv

In this second chapter you got familiar with Directly-Follows Graphs. This model is a very powerful tool to visualize different steps in the as-is processes to get a whole picture and spot bottlenecks.

After this second lecture you should now be familiar with structure and features of DFGs and how to create them in the Celonis EMS by uploading the data and customizing your view of the process. Furthermore, now you should be able to add different variants and apply filtering techniques.

Check your knowledge in the exercise section and see if you have understood both the theory and hands-on aspects of this lecture!

Lecture 2: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in an applied setting and transfer your knowledge to other use cases. You need to have completed all Knowledge Check sections in order to obtain your certificate!

Task 1: Create a Directly-Follows Graph.

You are given the same Purchase to Pay Event Log from Lecture 1 in fig. 1. This time you are provided with more information on how often the different variants occur in the total event log (= 12,659 cases).

Task 1.1. Based on the event log above, showing variants and frecquencies, have a closer look at the two most common variants. Based on the two variants, create a DFG in your notebook or a piece of paper. Tick the right statements for the DFG based on the two variants.

• The DFG starts with Create Purchase Order.

• The DFG starts with Create Purchase Requisition Item.

• Goods Receipt is followed by Scan Invoice in 2,962 cases.

• Goods Receipt is followed by Scan Invoice in 10,978 cases.

Task 1.2. Based on the event log above, showing variants and frequencies, create a DFG in your notebook or a piece of paper. Match the activities sequence with the respective number of cases on the arc between the activities.

problem paper link

Task 1.3. Look at the P2P analysis in Celonis. How many % of cases are described by how many % of all variants?

• 82% of all cases described by 1% of all variants

• 82% of all cases described by 7% of all variants

• 82% of all cases described by 2% of all variants

Which of the statements are true for arc-based filtering on DFGs?

• The final Directly- Follows Graph remains stable to any removal of activites.

• Hidden activities influence Directly-Follows frequency.

• Numbers do not need to add up after removing arcs.

Task 1.4. Look at the BPMN model below in fig. 2. In your notebook or piece of paper, draw a DFG as close to the BPMN model as possible. How many traces are possible?

• 3

• 4

• 6

04. Discover Sophisticated Process Models

Process Discovery: Higher-Level Process Models

Gain a deeper understanding of more sophisticated process discovery.

• WHAT’S INCLUDED

  • Self-paced learning. Take courses at your own speed.

  • Learn when it suits you with round-the-clock access.

  • Get enabled fast with laser-focused, goal-based training.

Lecture 3: Process Discovery (2/2) - Learning Higher-Level Process Models

Welcome to the third lecture unit and second part of the Process Discovery chapter, where we will explore more advanced Process Mining models. In the first part of this chapter you have learned that DFGs are a very powerful tool to show frequencies and reconstruct processes including bottlenecks. However, as you may have noticed, this process model also has some drawbacks, in particular when it comes to very messy data sets and tangled up process flows e.g. through concurrencies. And this is where higher-level process models come into play, and you will get familiar with them in this lecture!

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Let’s take our next step in your Process Mining journey now!

Lecture 3: Higher-Level Process Models

“If you are working with Process Mining tools, even if you don’t see Petri Nets, whenever you look at a BPMN model

• in the background the Petri Nets are there.”

Sometimes DFGs are not Enough!

The goal of the third lecture is to give you a better and deeper understanding of Process Discovery. As the data in the real world gets more complex, it is getting more difficult to represent it in a right way. In the focus of this lecture are Petri Nets and BPMN models, which are capable of dealing with concurrencies, choices, and loops. Using the real-world examples, Prof. Dr. Wil van der Aalst will also explain to you different ways to visualize the same data with different discovery models.

This lecture consists of a main part dedicated to Petri Nets and BPMN models and additional learning materials if you want to understand inductive mining at a deeper level. In the main lecture of this course you will learn:

• what a concurrency is and why DFGs cannot capture it
• what a Petri Net model is and how it functions
• the difference between bottom-up and top-down discovery approaches.

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

Additional Learning Material: What is Inductive Mining?

If you want to dive deeper into Process Discovery, have a look at the extra part of this lecture where Wil van der Aalst explains the basic principles behind inductive mining. Here you will learn more about different techniques to cut a DFG that help to analyse different subblocks of the model.

The third lecture has given you a deeper insight into Process Discovery and other process representation tools. In the focus of this chapter were concurrencies and how different process models can deal with them. You are invited to practice working with more complex data sets within Celonis. Visit the Celonis hands-on illustrations now to visualize the comples data with concurrencies yourself!

Lecture 3: Celonis Hands-On

“BPMN is the industry standard for process modelling. The unique selling point behind BPMN is the possibility for gateways which can reflect or or and relationships for choices and concurrency respectively”.

Other Process Representations in Celonis

In reality, business processes are hardly ever straightforward. Think back to our opening question about how packages get to your door so quickly. If every step happened just one at a time, there is no way you would receive your package in two days.

That’s why process discovery is so important. We need to figure out how things are actually happening so we can see the big picture. In Wil’s lecture we added a new component to the mix to account for this- what happens if processes don’t happen in a sequential or specified order? And as we very well know, they often don’t.

Enter the term concurrency. We need to be a little careful when we describe concurrency in relation to process mining. The term stems from the word concurrent, meaning to happen at the same time; however, in this context, concurrency means that no matter in what order a process is executed, even simultaneously, the end result will still be the same.

Think of this like making a strawberry cake versus making a strawberry smoothie. When baking a cake, all the steps need to be carried out in a specific order to achieve the same result- like how we can represent models with our directly follows graphs. However, with a smoothie, it doesn’t matter how you add things to the blender, you will alway still have a smoothie- like our more complex representation with petri nets. In this session, we will dive even deeper into process discovery in Celonis.

Before jumping into the practical part by Celonis, make sure that your Training environment is all set up:

Accessing your Celonis Training Team.pdf

In this lecture, we will work with the following data sets. Download them before watching the video:

concurrency-matters.xlsx

concurrency-matters-skips.xlsx

In the third chapter you have gotten familiar with different approaches of data visualization. Moreover, you have learned what a petri net is and how it can cope with concurrencies, choices, and loops. To test your knowledge on this topic, continue to the Knowledge check part of this chapter!

Lecture 3: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in an applied setting and transfer your knowledge to other use cases. You need to have completed all Knowledge Check sections in order to obtain your certificate!

Question 1: Which BPMN model was created using the inductive Process Mining algorithm?

Take a look at the Concurrency Matters dataset in the Celonis software. As you’ve just learnt in the Celonis lecture, Variant explorer cannot capture concurrences in the processes. For this, we want to have a look at the BPMN model. Take a look at the following models:

1

2

3

Which of the BPMN models corresponds to the variant explorer of the Concurrency Matters dataset?

• 1

• 2

• 3

Question 2: Understanding Higher level process models

Why are Directly Follows Graphs (DFGs) incapable of capturing the complexity of some real world processes?

• DFGs get tangled up very soon

• DFGs cannot capture parallel activities

• DFGs cannot measure timestamps correctly

Question 3: Apply your skills in Celonis

Upload the Concurrency Matters data set into your own training license. If you need help on how to do this you can revisit Lecture 2. Look at the process both in the BPMN model and variant explorer. How many connections can you count for each of them looking at whole process?

• 12 in Variant Explorer (DFG) and 8 in BPMN Model

• 15 in Variant Explorer (DFG) and 5 in BPMN Model

• 30 in Variant Explorer (DFG) and 10 in BPMN Model

05. Experience the Power of Conformance Checking

REQUIREDConformance Checking

Evaluate the “to-be” and “as-is” worlds.

• WHAT’S INCLUDED
  • Self-paced learning. Take courses at your own speed.
  • Learn when it suits you with round-the-clock access.
  • Get enabled fast with laser-focused, goal-based training.

Lecture 4: Conformance Checking

You are now half way through your Process Mining journey - congratulations! In this fourth lecture we turn to another important topic, Conformance Checking. It is a powerful methodology to see whether the real as-is process matches the desired behavior. Conformance Checking is of high importance for organisations that want to make sure that their processes are executed as planned.

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Let’s take our next step in the Process Mining Journey now!

“In conformance checking, we want to compare the model behavior with the observed behavior. This is highly relevant if an organisation wants to make sure that processes are executed as planned.”

Is the Process Executed as Planned?

In real life, things don’t always run according to plan. But for businesses and organizations it means losing control over processes and most importantly - money. If we look at the Directly-Follows Graphs and other process models, that we learned in the previous lectures, it can be very difficult to spot deviations at first glance. So far we have focused on Process Discovery. In Process Discovery you try to reconstruct the as-is process and spot first patterns all through your own research. With hundreds of process variants and tons of master data to analyze this can take a lot of time. Conformance Checking allows you to compare a normative process model, showing you the ideal process flow with your as-is process. It automatically detects violations that keep you from being in your ideal world.

In this lecture, Wil van der Aalst will explain how Conformance Checking can help organizations to observe where a desired model and real data deviate from each other. After watching this lecture you will be able to compare reality with a normative model with the help of three different approaches: Alignments, Footprint matrix and Token-based replay.

You will learn:

• the basics of Conformance Checking
• how to compute alignments and what their drawbacks are
• how to compare footprint matrices to spot deviations
• what a token-based replay is and how to analyse it.

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

In this lecture you have learned that there are three different approaches in Conformance Checking: Alignments, comparing footprints and token-based replay. A nice analogy that tells the difference between token-replay and alignments is searching for a particular place (e.g., a restaurant) in a city: in token-replay, you decide the direction to take just by looking at what you see. With alignments, you take your mobile phone and look at Google Maps, which will tell the optimal route (but pays the price of connecting to a GPS, download the city map, etc. …) (Josep Carmona, Associate Professor at Universitat Politècnica de Catalunya ).

By now you should be familiar with the drawbacks and advantages of all the methods and know how to create them. Now, it is time to see how Conformance Cheking is realized in the Celonis software! Let’s move on to the next practical lecture by Celonis to explore this is in application.

Lecture 4: Celonis Hands-On

“Maybe you stood in the mirror and practiced the perfect delivery of the question “tell me about yourself” or mapped out the expected direction of the talk. After the interview happened, you might have thought back to your original ideas of the conversation and considered whether the actual outcome met your expectations.”

Conformance Checking in Celonis

In the past lectures we talked about different methods to turn process data into process models. The “Conformance Checking” allows you to design, mine or upload a normative process model in the form of a BPMN model, and automatically compare it to the as-is process discovered in the data. Subsequently, you will get an overview of all deviations from the should-be process flow. Conformance Checking also allows for an automated root cause analysis which will be a subsequent step in our journey.

Celonis allows for several ways to come up with your normative process model. You can design a standard process from scratch in Celonis using the integrated BPMN process modeller, select an existing BPMN model from the Celonis Process Repository, or even mine a process model directly from the variant explorer with the help of inductive mining techniques.

Process violations are automatically flagged by Celonis. We’ve already discovered this undesired activity of changing a price when we looked at the process in the form of a Directly-Follows Graph with the Variant Explorer. The conformance checker provides us with a whole list of violations that keep us from conforming with the should-be process and hence require action and attention.

In the background the event log is replayed on the Petri Net, to help compare the as-is and should-be model, understand potential root causes and identify correlations between a selected conformance violation and associated metadata of the corresponding purchase order items. Metadata can be a vendor name, a material group or a specific plant just to name a few. We will also talk about the influence of metadata on the process further in Lecture 6.

Before jumping into the practical part by Celonis, make sure that your Training environment is all set up:

Accessing your Celonis Training Team.pdf

In this lecture, we will work with the following data set. Download it before watching the video:

concurrency_matters_dev.xlsx

You are encouraged to study the demo data yourself in Celonis. View the example from the Lecture video here!

DEMO NOW!

Conformance Checking can be very handy in real world applications when Process Discovery can still require a lot of human interaction. Conformance Checking then comes in as a fast way to spot deviations. Together with Process Discovery and Enhancement it is an essential pillar of Process Mining theory and a key diagnostics tool.

Lecture 4: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in an applied setting and transfer your knowledge to other use cases. You need to have completed all Knowledge Check sections in order to obtain your certificate!

Task 1. Which techniques exist for Conformance Checking? Tick the right boxes.

[x] Alignments

[x] Footprint comparison

[x] Token-based replay

Task 2. Have a look at the Petri Net (model) on the left and the event log variants with frequencies (observed) on the right. How many cases are conformant with the model?

• 12,659

• 10,978

• 1,681

Task 3. Have a look at the BPMN (model) on the left and the event log variants with frequencies (observed) on the right. How many cases are conformant with the model?

• 12,659

• 10,978

• 1,681

06. Analyze Process Data

REQUIREDGetting the Data from Source Systems

Understand where process data comes from and potential complications.

• WHAT’S INCLUDED
  • Self-paced learning. Take courses at your own speed.
  • Learn when it suits you with round-the-clock access.
  • Get enabled fast with laser-focused, goal-based training.

Lecture 5: Getting the Data

Process Mining allows for analysis of millions of cases, and therefore is a big data technology. However, huge amounts of data also leave room for errors and data issues are among the biggest threats to a successful Process Mining implementation. Incorrect time stamps, currency conversion problems, or gaps in the data can cause a headache for Process Miners.

The focus of this lecture is on understanding the data collection mechanism behind process mining and how to go from the raw data to a process model to an analysis.

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Let’s take our next step in the Process Mining Journey now!

“It is a big challenge to extract data from the information systems, and if you start Process Mining from the very beginning it might take up to 80% of your time.”

Looking for Quality Data

It is hard to understand Process Mining without first understanding Data Mining. The data underlying our Process Mining analysis is vital and so is the work around preparing the data. As the above quote from Wil rightfully points, out getting the data can be tricky. It takes up a lot of time and there are typical errors that can occur from switched timestamps to gaps in the data.

Wil has dedicated this lecture to explaining the nitty-gritty work of mining and preparing our Process Mining data. More specifically, you will learn:

• What are the challenges behind data extraction
• How to build a case-identifier with a given dataset in different environments
• The Transactional Model for Activities
• Atomic activities versus activities with a duration
• Possible problems from deficiency to convergence to divergence

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

In this lecture you have learned about a more refined view to work with data in Process Mining. Let’s jump into practice! Join the next practical session by Celonis where you will learn how to upload data and build complex data models in Celonis.

Lecture 5: Celonis Hands-On

“Remember however that in practice data sets are normally astronomically large and have to be sourced from a number of different systems”

Data Upload and Building Data Models in Celonis

We already took a look at Celonis’ interface in the previous lectures, but we’ve yet to examine how the data we are analyzing is uploaded into the system in the first place. In Wil’s lecture for this unit, you learned about basic principles of finding data and assigning case identifiers. Remember, however, that in practice these data sets are normally astronomically large and have to be sourced from a number of different systems. Now, we’re going to see how Celonis helps companies effortlessly find and extract data for simplified navigation. We’ll walk through how data files are uploaded and construct our very own dashboard from scratch.

Before jumping into the practical part by Celonis, make sure that your Training environment is all set up:

Accessing your Celonis Training Team.pdf

In this lecture, we will work with the following data sets. Download them before watching the video:

P2P-Activities.csv

P2P-Cases.csv

You should be a pro in data-related questions by now! Data upload is no one-off task anymore. Instead of analyzing historical data only, Celonis allows for real-time data upload to monitor a process in real-time and also set preventive measures in real time. We will take this up further in lecture 6 when we will also address process prediction.

Ready for the next test of your Process Mining knowledge? Then continue to the Knowledge Check we have prepared for you.

Lecture 5: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in an applied setting and transfer your knowledge to other use cases. You need to have completed all Knowledge Check sections in order to obtain your certificate!

Task 1. Looking at a more refined view of event data, we have 3 concepts: process, case and event. Events may have many attributes. Sort the information on top into the right attribute category.

problem paper link

Task 2. What are the challenges when extracting event data? Tick the right boxes.

• Finding the data

• Extracting the data

• Understand the data

• Migrating data to a repository

• Quality problems

• Scalability

• Data security and regulations

• Too good record keeping at companies

Task 3. One of the goals of the IEEE Task Force on Process Mining is to promote the use of Process Mining techniques and tools. An important aspect of that is the existence of common and widely-accepted standards, one of it being the data format for event logs. XES is an XML-based format, and its name is an acronym for ____ ___ _____.

Task 4. There are multiple problems that we encounter when we try to convert an object-centered event log into a conventional event log. Match the right term descriptions!

problem paper link

Task 5. A data-driven perspective on business process management problems has huge advantages in terms of covering complexity as well as the effort and degree of detailedness for gathering information and evidence. Match these other frequently used techniques in Business Process Management and Consulting to gain insights into processes with their biggest shortcomings.

problem paper link

07. Chart Comparisons and Predictions

Comparative and Predictive Process Mining

Spot patterns and predict the future with Process Mining.

• WHAT’S INCLUDED
  • Self-paced learning. Take courses at your own speed.
  • Learn when it suits you with round-the-clock access.
  • Get enabled fast with laser-focused, goal-based training.

Lecture 6: Comparative and Predictive Process Mining

So far we have just looked at processes from a historical viewpoint, analyzing past data. However, Process Mining also has a huge application in analyzing real-time data and making predictions for the future! We can leverage the power of Machine Learning to create predictive models based on our process discovered. Closed cases from historical data become the base for real-time prediction and pattern recognition. In the sixth lecture you will learn more about comparative and predictive Process Mining.

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Let’s take our next step in the Process Mining Journey now!

“If we are able to build a model to compare, that is enriched with all these additional perspectives, we can use the very same model to make predictions.”

Let’s See into the Future: Using Process Mining as a Crystal Ball

In the previous lectures we learned how to use event data to discover process models. But what if we could do even more with our discoveries? By extending process models with additional data, we can use them to make comparisons and predictions! This can be crucial in business: just imagine getting insights from the data about recurring patterns behind bottlenecks and the factors that cause them.

Imagine you can predict when e.g. order deliveries will be late or which vendors are likely to cause manual rework and hinder automation For these kind of problems, the integration of Machine Learning techniques into Process Mining is a very powerful approach. In this lecture, Wil van der Aalst will focus on Comparative and Predictive Process Mining and their respective link to Machine Learning and other prediction mechanisms.

More specifically, you will learn:

• the basic principles of both Comparative and Predictive Process Mining
• how to extend process models with additional attributes
• how Neural Networks can help us make predictions

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

By now you are familiar with the basic principles of Comparative and Predictive Process Mining, their link to Machine Learning and Neural Networks and how they can help humans make decisions.

Now, it is time to dive deep into the comparative and predictive Process Mining in the Celonis software! Let’s move on to the next Practical Lecture by Celonis!

Lecture 6: Celonis Hands-On

“In the past, Process Mining was only used to analyze historical data. However, real time data loads and being able to analyze process flows as they happen now opens up completely new horizons”

Comparative and Predictive Process Mining in Celonis

Wil’s lecture was a great representation of the importance of Machine Learning and AI in the future of Process Mining and Execution Management. We’re going to take a look at how Celonis facilitates predictive analytics so companies can take proactive steps to improve their processes. Predictive analytics is a little bit like becoming a detective. We can take a look at evidence, in our case historical and even real-time data, to find patterns in processes and furthermore make predictions about how these processes may act in the future.

If you’re Sherlock Holmes in this situation, then Celonis is akin to John Watson. Not only does Celonis help you automatically recognize patterns and regularities in process data but Celonis proactively brings issues to the forefront of your attention and even suggests fact-based recommendations for your next-best-actions to optimize your desired business outcome.

In this sixth lecture you have learned more about future-oriented Process Minings. By now you also know how Machine Learning can be integrated in Celonis and how it can help businesses make smart decisions. Celonis even has its own ML algorithms specifically trained on process data which help you to create a predictive model. Are you ready to test your knowledge?

Lecture 6: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in an applied setting and transfer your knowledge to other use cases. You need to have completed all Knowledge Check sections in order to obtain your certificate!

Task 1. Have a look below at the two Directly-Follows Graphs for the rental car process comparing two different customer groups. What differences can you spot in behavior and performance? Tick the correct statements.

• The validation of information takes less time for business customers.

• The validation of informantion takes more time for business customers.

• In the majority of cases business customers set up a new account.

• In the majority of cases private customers set up a new account.

• Private customers need to reinitiate the process more often than business customers.

Task 2. With extended event data we can do different types of analyses to discover our process. Match the analysis type to the right question.

problem paper link

Task 3. What term are we looking for?

Machine learning algorithms that use this generally do not need to be programmed.

It translates input features into one or more target features.

For a long time it was not very successful and classical techniques like decision trees and regressions were preferred.

problem paper link

Task 4.

With the help of the results of the feature selection algorithm, we can select the features that we want to use for our model. We distinguish between two features types. __ features do not change throughout a case and always remain the same. Examples could be Customer or Material Number. Dynamic features are features that change over time and need to be updated throughout a case. When building a training set they need to be treated differently since they should not have the last state of a case but the state at a given time in the process. Examples are counts of how often a certain activity already occurred in the process.

problem paper link

08. Current Process Mining Trends

Wrap-up Process Mining from Theory to Execution

Process Mining Quo Vadis. Learning about RPA and other trends.

• WHAT’S INCLUDED
  • Self-paced learning. Take courses at your own speed.
  • Learn when it suits you with round-the-clock access.
  • Get enabled fast with laser-focused, goal-based training.

Lecture 7: Closing & Outlook

Congratulations, you have made it to the last lecture of Process Mining: From Theory to Execution! In this lecture we are going to recap everything you have learned so far. Also, Wil will give us an outlook of how Process Mining has developed in the last years and current trends in the field of Process Mining & Automation. If you want to further expand your knowledge don’t forget to check out our Resources chapter at the end of this course. It includes e.g. a discount code to Wil’s world-famous book “Process Mining: Data Science in Action” (2016).

Just like all of the other lectures this lecture unit contains a theory lecture from Wil, hands-on illustrations in Celonis as well as a knowledge check quiz! Let’s take our final step in the Process Mining Journey now!

Lecture 7: Closing

“If you are looking for an operational friction, focus on the things that are infrequent and that you don’t want to happen.”

Wrapping Up

It’s time to recap everything we’ve learnt about Process Mining in the last six lectures! In the closing lecture, Wil will lead you once again through the different aspects of Process Mining and how they come into real world application. After this short overview, Wil will give you a quick outlook into Process & Automation which are said to be a match made in heaven. Process Mining helps to detect bottlenecks and monitor automation initiatives.

Let’s dive right into the last lecture with Wil!

Hint: We recommend taking notes and listening actively since there will be quiz questions testing your knowledge!

In this lecture you have repeated all the important topics of Process Mining and got a sneak peek into Automation. Now, you are ready for your last practical lecture with Celonis!

Lecture 7: Celonis Hands-On

“In Celonis, we can easily see our team’s process - exactly as it is occurring in our systems - and identify areas in which we should improve.”

The last Stop of our Process Mining Journey

Welcome to the final part of the Celonis illustrations of this theory to execution class. Today we want to spend some time talking about how Process Mining has advanced in the last few years and how Celonis has developed from its Process Mining core technology towards an Execution Management System. So far we have learned the ins and outs of Process Mining from Process Discovery to Process Conformance to Process Enhancement. In recent years, Process Enhancement has become increasingly important. Two key elements to implement lasting process improvements are automation and prediction. In Celonis these features are captured as part of the Celonis Studio, Execution Apps and Action Flows. These key features of Celonis also play a key role in the daily operations in the usage of the Execution Management System. The Celonis Studio and features that come with it are particularly important for the daily analysis of incoming real-time data. In order to act fast and prevent future bottlenecks, automation and prediction are essential tools.

We now want to look at these tools in more detail.

You have made it through 7 lecture units of this course now! It’s time to wrap. This last session gave you an outlook into current trends in the field with a particular focus on Automation. We hope that you enjoyed your Process Mining journey together with us. Process Mining is a huge field spanning from Process Discovery, to Conformance Checking and Process Enhancement. We hope you now have a better of what this field entails and also the maths behind it. Wil has taken your into the maths and theory behind this powerful field and together with Celonis you will have seen how this applies in state-of-the-art technology!

Before ending this journey and getting your Process Mining badge, proceed to your last knowledge check!

Lecture 7: Knowledge Check

Check your Knowledge!

Let’s see how much you have taken away from your lecture so far! Take a look at the questions below and try to figure out what the right answer might be. These questions are designed to be slight brainteaser so you can apply your theoretical knowledge in an applied setting and transfer your knowledge to other use cases. You need to have completed all Knowledge Check sections in order to obtain your certificate!

Task 1. What does the Pareto distribution tell us in relation to processes? Tick the correct statements.

• 20% of the cases, is explained by only 80% of variants.

• 80% of the cases, is explained by only 20% of variants.

• Infrequent variants often correspond to the desired process.

• Infrequent variants often correspond to problems in the process.

Task 2. Match the different topics that we have covered in the course with the respective contents.

problem paper link

Task 3. What is the basic idea of RPA (= Robotic Process Automation)?

• Replacing the system in the back-end by an automated system

• Replacing the human in the front-end doing repetitive tasks by an automated system

• Replacing the human in the back-end solving complex tasks by an automated system

Task 4. In this course you have learnt a lot about the theory of Process Mining and the application within the Celonis Execution Management Systems (EMS). Which of the following statements should you take home with you?

• The Celonis EMS is based on Process Mining but provides a full cloud platform for Process Analysis and Execution.

• The Celonis EMS gives us all the relevant information that we do not need to question.

• The theory of Process Mining helps us to understand what is going on behind the scenes of the system and better evaluate what the system presents us with.

09. Your Feedback

ELECTIVEFeedback on the Process Mining: From Theory To Execution Track

Tell us what you thought about this training track!

• WHAT’S INCLUDED
  • Self-paced learning. Take courses at your own speed.
  • Learn when it suits you with round-the-clock access.
  • Get enabled fast with laser-focused, goal-based training.