Getting to know the process
1. Getting to know the process
So let’s quickly sum up the different steps we just when through in creating the process data, and take a closer look at this.2. Identify data sources
Firstly, you have to identify the different sources of data related to the process under consideration. These might be very different formats, and have different data structures.3. Reshape them
Secondly, make sure they are all transformed so that each row is an event. This will often require some data reshaping using tidyr and dplyr.4. Harmonize them
Thirdly, verify that each of the datasets has been given uniform variable names and levels. This step will generally require renaming some variable, recoding some level, or maybe adding some missing variables. While the previous steps focussed on dividing the data in row, or events, this step focusses on appropriately formatting the data related to each events, i.e. the columns.5. Create event data
Finally, we can put the data together, and provide it to the eventlog constructor function. Now, we are ready to analyze it.6. Inspect event data
Before we start with in depth analysis or with answering research questions, it is recommended to have a high level understanding of our data. Inspecting the otc event log we see that it has 37 different activities and describes 3651 cases, which took place between the January 1st of 2017 and April 22nd of 2018.7. Activity presence
Looking at the activity presence, we see that about 10 activities occur in every single trace, therefore making up the backbone of the process. The remaining 27 activities have differ in their presence, from relatively often to very exceptional.8. Trace lengths
We can also look at the number of activity instances per case. The majority of cases has a trace length between 13 and 21 activities, although the minimum number is as low as 3, and the maximum goes up to 30 activity instances.9. Entry points
Further analysis shows us that there are two starting points in the process: receiving a request for quotation or a sales order.10. Exit points
While most cases logically end with receiving the payment, a minority end otherwise: the request is declined, or the quotation is cancelled. A small proportion of cases are send to a credit collection agency, while other cases end with the creation of a credit note or the sending of an invoice. The latter two activities do seem inappropriate ending points, and might form a starting point for some analyses. Before we do so, let’s take a look at the overall structured of the process, and its most common variants. We’ll do that in the next exercises.11. Let's practice!
Let's start!Create Your Free Account
or
By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.