We are pleased to introduce the new version of LANA Process Mining*. With version 4.10.0 we support Python for custom dashboards. In addition, we have improved usability with model legends and error messages for the model upload.
* The release for on-premises will be about 2 weeks later.
Seeing and understanding
The new legends in the discovered and target models allow you to understand what is being displayed at a glance. In the upper right corner you will find the legend for the selected size and color metric. In the lower left corner, you will find a legend that explains the coloring in more detail. Both legends can be expanded and collapsed by clicking on them.
Improved model upload
The model upload experience has been significantly improved. If a model cannot be uploaded, the user now receives a message. This message summarizes why the model cannot be uploaded to LANA so that you, the user, have the possibility to improve the model.
Other improvements and bug fixes:
- Charts on the Dashboard page used too much memory
- Shiny charts could not be created on the dashboard page
- Rendering of path labels in the actual model for throughput as a color metric was incorrect
- The target model was not connected to the event log under certain conditions
- The distribution analysis on the Insights page failed because the event log data only contained attribute values with zeros
- The conformance chart on the statistics page could not be downloaded as a PNG, PDF or SVG file
- It was not possible to create a filter from the “Most important categorical attributes” list on the “Insights” page
- Search on the “Variants” page did not work as intended
- Roles next to the RAnalyst, AdvancedAnalyst, SuperAdmin were able to upload dashboard sources (Shiny and Python)
- The title of the case duration chart on the “Statistics” page was not translated to the user’s chosen language
- When uploading a log with a different time zone than the user’s, the front end increased/decreased the start and end time of the throughput metric
- Initial loading of the dashboard resulted in too small charts for actual models