You can as usual download it from www.knime.com and install it free of charge since KNIME is still and will stay an open source software.
Having two releases per year might lead to many update headackes. But I must admit that the developers at KNIME have done an incredibly good work at keeping the back-compatibility with all previousoly existing nodes and frameworks. With each new version my many old workflows are still working and, besides the occasional message "This node has been deprecated", without any hiccups. But moving back to the lates KNIME 2.9 version, there are a few changes that KNIME users will appreciate.
First of all, there is a whole new family of rule based nodes. The Rule Engine node has a new dress, following the same User Interface as the String Manipulation node and later as the Math Formula node. This is a big step in uniforming the user interfaces. And not only! The new Rule Engine node allows now for a more extensive editing of rules including flow variables as well as data columns. In addition, the new Ruleset Editor node allows you to define a rule set and to export it as a PMML model. Note that a PMML rule set model cannot deal with flow variables, though.
Two new nodes, also based on the new rule engine framework, allow for filtering and splitting rows in the input data set on the basis of the edited rule set: the Rule-based Row Filter and Rule-based Row Splitter nodes.
A new column selection framework has been introduced. This new framework allows for the selection of data columns manually (like before), by type, and by column name, where the column name pattern can be a RegEx or a wilcard expression. This new framework applies to the Column Filter node of course, but also to all other nodes where a column selection is required.
Data Mining Nodes
The output of the Linear Regression node now includes the linear regression coefficients.
For the time series analysts or for those just modelling numerical data, the new Numerical Scorer node might be of interest. This node calculates the error between the numerical predicted output and the numerical desired output, using a number of different distance measures.
And, finally, for the Weka lovers, Weka 3.7 has been integrated with this latest release.
New nodes and new options have also been introduced to monitor and speed up workflow execution. Here I would like to just note the new Email node. With this node, your workflow can send you email updates about the workflow execution up to that point. This can be a very useful node for long executing workflows.
And more ...
There are many other new nodes and features, with less impact than the ones already described here. If you are interested, you can check the full list of changes at the "What is new in KNIME 2.9" web site http://tech.knime.org/whats-new-in-knime-29 .