Preface
This document is intended to guide you through the different steps involved in developing the Sacumen process mining connector.
Requirements
Before beginning this guide, make sure that the processes and data sources you choose are appropriate for your process mining implementation. Process mining requires:
- An action that describes the process event that is being performed.
- Timestamp showing when each step took place
- An identifier that combines all events into one payment.
Data sample
Many of these steps are part of the data conversion process. This conversion is necessary because Process Mining needs to be able to display the data in the form of event logs. This format is called a data model. The data model defines the tables and attributes that should be part of the output data set generated by the connector. However, most source systems do not provide this data format. So you need to transform the data.
Union
To make data transformation easy, Process Mining has created several connectors that you can use as templates to load, cleanse, and transform data. These precompiled connectors are specific to some systems and processes, e.g. B. SAP Connector Development for Purchase to Pay Discovery Accelerator.
When starting a new project, you should check if you have the Sacumen connectors ready. In the simplest case, the input data fits directly into one of the sockets and does not require a special socket. If none of these options apply, you must create the connector yourself.
Create a connector
The purpose of the connector is to create a dataset that fits your data model. The most common process data mining model consists of case tables and event logs. The alternative data model consists of several entity tables and event tables. Regardless of the data model, it is recommended to configure the first step of the connector conversion as shown below.
Support model
In SQL, not all conversions can be calculated on the same table. This may be due to an aggregate or function that cannot be represented by a single select expression. To do this, you can create a support table. You can use the model for multiple transformations by creating a maintenance table in the database. You can also add supported transformations to an existing table as a preprocessing query if you don’t need to reuse the model. It’s a balance between creating a child table and preprocessing the query for a readable conversion.
You can group them into different subdirectories to distinguish supported conversions from others. The default name of this group should be support models, but you can use a more descriptive name.
Naming convention
After completing the steps above, create a connector base folder structure with the following naming convention:
1_author
2_object
3_Events
4_Event logs
5_logika_business
Within each of these main folders there may be a promoters models folder or some other group that will help you better understand the transformation.
The model only needs to depend on the model until they are in the same root group. Therefore, you cannot define a table in 3. Events associated with the Event logs table.
About this guide
This guide provides step-by-step instructions, conceptual explanations, and tips and tricks on how to practice creating these different sections. This guide is general and can be applied to any process and system. To follow this guide correctly, make sure you are familiar with the inputs and the underlying model.