Sympathy for Data v1.4.2 available!


Sympathy for Data version 1.4.2 offers reliable support for Python 3, improved performance and several new features, such as, a machine learning library, improved startup time and many small improvements to the standard node library.

Nodes and flows from 1.3.0 should be compatible with 1.4.2.


Download Sympathy for Data built for Windows:


Node/plugin changes:

  • Improved performance of Select columns in Table in cases when lots of columns are discarded
  • Added the node Periodic Sequence Split Table that can split up a Table into a Tables list where each element holds one periodic event
  • Support for creating masked values in Lookup Table and Pivot Table. Pivot node now works with any data type
  • Improved performance in all VJoin Tables/ADAFs nodes with a single list input, in the case when the input list contains a single element
  • Optimization and new option for HJoin Tables with different number of rows analogous to VJoin Tables
  • Optional creation of masked array in Ensure columns in Tables with Table
  • Harmonized quoting for CSV importers and exporters
  • Chunked MDF writing to improve performance when exporting large Rasters
  • Extended VJoin Tables with more options for controlling how to handle missing index

Machine learning:

New machine learning nodes based on scikit-learn. Features include:

See also the machine learning examples from the install path of the Sympathy node library.


  • It is now possible to enter a minimum version for a workflow, in flow properties. Bear in mind that this feature is not very useful until it has existed for a few versions
  • Possibility to jump from an error message to the node/flow that caused the error
  • Some default workflow variables have been removed. Only SY_FLOW_FILEPATH and SY_FLOW_DIR remain
  • Lambdas can be configured to show input ports that can be used to perform configuration against data. See Lambda function
  • Improved performance of the Figure type in some situations
  • Redesigned sympathy.api.dtypes, this API should be stable
  • Configurable node ports can now be added and removed when the nodes are connected. See Ports
  • Made it possible to build GUI:s from group parameters which includes children.
  • Improved handling of flows and libraries in non-ascii paths
  • Reduced the maximum number of worker processes, used by default, to 4