Enhancements to Modern UI/UX
KNIME has been investing heavily into improving the user experience recognizing that ease of use and ease of onboarding is critical to upskilling. To that effect, new enhancements to the UI/UX include:
- The workflow tab bar to work with multiple workflows in multiple tabs, for instance, to copy and paste nodes from one workflow to another
- The ability to add dynamic ports directly to natives nodes, similar to components and metanodes. Simply drop the ports on the “+” icon
- Improved onboarding and usability by getting the recommendation for the 12 most commonly used nodes, based on workflow coach data, when a port is dragged out of a node. Selecting the desired node connects it directly. The workflow coach recommends which nodes to use when and in what order based on anonymous KNIME usage data
- Table view for basic filtering, basic search, and resizing columns. This enables performant scrolling through large tables with lazy loading and virtual scrolling
This new Geospatial Extension provides the ability to carry out geospatial analysis within KNIME. In its first version, it will support the most common vector data types such as points, lines, polygons, and collections of those. It is a joint development by the Center for Geographic Analysis from Harvard and KNIME bringing together top geospatial and KNIME development expertise. It is a trusted community contribution and we will continue to improve it further.
Major functionality includes:
- Reading and writing different geospatial files such as shapefiles or GeoJSON files
- Performing spatial calculations such as computing the area of a polygon, computing distances, performing spatial joins, and other spatial manipulations
- Spatial Transformation nodes that convert points to lines or harmonize the projection of two different datasets
- Built-in visualization nodes for interactive views that can be included in KNIME Data Apps, and richly configurable static maps for presentations or reports
- Nodes to easily access publicly available data such as Open Street Map or the US Census data
As always your feedback is very important to grow this extension further, please feel free to contact us via the KNIME Forum or Github. In addition to the Geospatial Analytics Extension we will publish many workflows for the most common geospatial use cases in the KNIME Community Hub so everyone including non geospatial experts can hit the ground running.
Improved Integration Between KNIME and Python
With KNIME AP 4.7 the new Python integration is now ready for your production environments. Its most notable improvements are much faster data transfers between KNIME and Python, a better Python API, and a bundled Python environment to get you started without having to install Python yourself.
In terms of scripting functionality, this release brings:
- Enhanced script API and type support especially for date and time formats and chemical types
- Improved RowID handling and RDKit support for molecules, fingerprints, and reactions
- Shipping of more recent versions of Pandas, PyArrow, scikit-learn and the addition of OpenPyxl
- Simplified custom Conda environment creation, Conda packages available at knime :: Anaconda.org
- Support for native M1 Mac Python builds
Please note that the previous “KNIME Python Integration” is now called “KNIME Python 2 Integration (legacy)” and should from now on only be used if Python 2.7 support is required. Similarly, “KNIME Python Integration (Labs)” now changes to “KNIME Python Integration” and loses the “Labs” suffix, same goes for the nodes and preference pages. See updated documentation here and here.
New Python View Node
With the new Python integration, we are excited to also provide the new Python View node supporting HTML based views which allow for interactive plots. The new view can also interact with other plots by synchronizing the selected data points. Take a look at sample workflows that use the Python View node here.
Enhancements to KNIME Python Extension Development (Labs)
Developing KNIME nodes in pure Python is now easier with improvements such as:
- Managing dependencies on other features and KNIME Analytics Platform version
- Allowing deprecation of pure-Python nodes
- Support for pip packages in bundled Conda environments
- New Enum parameter
- Versioning support for node parameters (see Docs)
- Markdown of node descriptions
New Visualization Nodes
You get seven brand new visualization nodes - box plot, density plot, heatmap, histogram, pie chart, stacked area chart, and statistics. Here are some sample workflows that use these new visualization nodes.
Stacked Area Chart
More Efficient Connectivity/File Handling
The new DB Row Manipulator node that allows users to write their own SQL statements to efficiently manipulate data within the database. Common use cases for a custom SQL statement are casting data such as geospatial data before sending it to the database.
Other additions include:
- Optimizing automatically generated SQL statements to improve execution performance for databases like SAP HANA
- Database metadata browser improved to not only show standard tables and views but also database specific ones such as external tables
- Added support for Azure AD application and service principals to the Microsoft Authentication node
- SAP Reader (Theobald Software) updated with support for the new dedicated KNIME destination in Xtract Universal
Connecting Multiple Workflows Independent of Workflow Location
Building complex solutions often requires multiple KNIME workflows and the ability to call one workflow from another is key to connect these workflows via an API. With this release, the various nodes to call one workflow from another have been adjusted to follow the same usage concept, independent of where the workflow itself lives: KNIME Server, KNIME Business Hub, or in your local workspace.
Two additional nodes Container Input (Raw HTTP) and Container Output (Raw HTTP) have been added to define generic REST APIs via KNIME workflows. These expand the application area beyond JSON-based REST APIs. For instance, workflows defining dynamic content on web pages, consumption and generation of additional data formats such as files and images and the deployment of workflows as WebHooks for integration into other systems.
New Utility Nodes
The Sorter and Top K Selector nodes provide an additional option to sort strings alphanumerically, thus supporting a more natural sorting of data. Often, data is enumerated by numbers such as “Row1” < “Row2” < “Row10.” The old, but still supported, lexicographic ordering would compare “Row10” < “Row2.” This sorting order is also applied when inspecting the data at a node’s outport in both the classic KNIME workbench and the modern UI preview.
Row To Column Header is a new node to extract a data row and convert it into the column header. How often have you read data from, e.g. an excel sheet but the header information is hidden somewhere “in the data.” This new node will help, simple type conversion included.
Native KNIME Builds for Apple Silicon Processors
KNIME Analytics Platform 4.7 is now also available with native product builds for Apple Silicon processors M1 and M2. In comparison to the Intel-only model, either on Intel chips or emulated on the new processor generation, these deliver superior performance, providing a more responsive user interface and far better execution performance. Users of Apple Silicon model should consider a fresh new install on this architecture.