Creating a Process Mining project

After installing and accessing IBM Process Mining, you need to create a project with data sources to begin analyzing your processes. IBM Process Mining supports multiple approaches to creating data sources, depending on your needs and available data.

You can use the following data sources in a IBM Process Mining:

Standard event log with process analysis

The standard approach involves uploading an event log that contains process execution data. This enables full process mining capabilities including process discovery, conformance checking, and performance analysis.

For detailed information about event log requirements, see Event log.

For more information on how to create a single-process project, see Creating a single-process project.

For more information on how to create a multi-process project, see Creating a multi-process project.

Object tables (dashboards and contextual data)
Object tables can add contextual data to your projects. By creating object tables, you can upload master data that is related to the context of your process (for example, data about vendors, materials, countries) that is separate from the event log. For more information, see Object tables.

You can also upload your master data as object tables without an event log to use IBM Process Mining as a business intelligence tool. This approach does not enable process analysis features but allows you to build dashboards and visualizations on your uploaded tables. For more information, see Using the Analytics page.

Simulated process from BPMN
If you do not have event log data, you can generate a synthetic event log from a BPMN (Business Process Model and Notation) model. This approach allows you to simulate process execution and analyze the results as if they were real process data. For more information, see Simulation.
External data sources

In addition to uploading files, you can connect to the following external data sources for automated data ingestion:

Data streams
Real-time integration through IBM Event Streams, Kafka, or IBM MQ. For more information, see Data stream.
Schedule sources
Automated retrieval from cloud storage including IBM Cloud Object Storage, AWS S3, or Azure Blob Storage. For more information, see Schedule source.
Process Apps
Pre-built connectors for extracting data from enterprise systems. Use the button on the main page of IBM Process Mining create a project from a process app. For more information, see Process Apps.

Creating a single-process project

Procedure

To create a process-mining project with your own .csv log files, do the following steps:

  1. Click Create a process-mining project on the home page.
  2. Type the name of the project.
  3. Choose one of the following options:
    • To link the project to an existing organization, click Use existing organization, and select the organization from the Organization name list.
    • To link the project to a new organization, click Create new organization, and type the name of the organization.
  4. In the Data source page, click Upload file and add a data source for the project.
    Note: You can select multiple data sources at the same time. They will be uploaded one after the other. If any of the chosen files is invalid, you will see an error message. All the other valid files will be uploaded despite the error.
  5. Optional: Add a backup file or a reference model.
  6. Click Edit data mapping and map your data source column according to the log data mapping requirements. For more information about event log creation requirements, see Event log.
  7. In the Configure the time format window, select the time format for your data source.
    Tip: You can choose a different time format for time-based mapping that is configured in the Map your data tab. To use this feature, clear the Use same time format for all dates checkbox, and select a different time format.
  8. Optional: In the Custom configuration window, select the required fields and format for all your custom fields from the Map your data tab.
  9. Click Create project.

You are now redirected to your newly created process. To see whether any of your CSV files have any warnings or errors, go back to the Data & Settings tab. For more information, see Managing error and warning messages.

Creating a multi-process project

A multi-process project is a project containing other processes from the same organization. An object-centric approach to data preparation must be used to create a multi-process project. Object-centric process mining is an approach that tracks multiple business objects (such as orders, invoices, deliveries, and customers) as they move through interconnected processes, rather than forcing everything into a single case-based view.

Object-centric process mining uses:

Event logs
Recorded activities with timestamps (e.g., "Order Created", "Delivery Shipped")
Object tables
Information about business objects (e.g., order details, customer information, product data)
Relationships
Relationships between objects (e.g., how order_id links orders to deliveries)

Before you begin

Before creating a multi-process project, do the following steps:

  1. Create at least two flat (single) projects in the same organization.
  2. Add object tables that will connect the processes. To learn more about creating object tables, see Object tables.
  3. Connect the processes through common columns in object tables. To learn more, see Table joins.

Procedure

Do the following steps to create a multi-process project:

  1. Click Create a process-mining project on the home page.
  2. Type the name of the project.
  3. Click Use existing organization, and select the organization that contains your processes from the Organization name drop down list.
  4. Set the Process type toggle from Single process to Multiple processes.
  5. Click Create Process.
  6. Select up to 5 projects (processes) that you want to include in your multiprocess project. The compatibility of the event logs is automatically checked, as the algorithm automatically detects valid paths between event logs, validates completeness of primary keys, and ensures correct correspondence between the case IDs of each process. If your processes are incompatible IBM Process Mining returns and error.
  7. Specify the order of the selected processes. Multiple combinations of processes and the corresponding join paths can be used to generate an object-centric event log.
    Tip: The order of processes should generally follow the temporal sequence of events across event logs. You should start with the process that contains events which occur earliest in the overall workflow and proceed with the other processes based on the expected business use case.
  8. Optional: Choose the color of each process. Colors will be visible on the visualized process model.
  9. Click Visualize.

Managing error and warning messages

After you have uploaded your project's data source (or sources), completed data mapping, and visualized the process, you can go back to the Data source tab and see whether your files have any error or warning messages:
  • A Warning status informs you that one or more rows of the file has been discarded from computation. The status appears in the following scenarios:
    • If there is no carriage return (empty line) after the last record of your file.
    • If a string column is too long.
    Hover over the file status to see which warning you have, and on which line-column. The warnings inform you that the process model and statistics might not be computed correctly (for example, if there is no carriage return, the last line of you CSV file might not be computed, meaning your model and statistics are lacking some data-rows).
    Tip: If the number of errors or warnings is more than 3, click the download button to see the whole list.
  • A Failed status informs you that the whole file was discarded. The status appears in the following scenarios:
    • If 100 or more warnings are found (for example, 1 "no carriage return at the end of file" warning and 99 "string too long" warnings, or 100 "string too long" warnings).
    • If at least one error is found. Errors are more serious that warnings. Hover over the file status to see the error you have.
If the number of errors or warnings is more than 3, click the download button to see the whole list.