News

KNIME 2.12. The Facts. The Features. New Functionality.

August 6, 2015 — by Bernd Wiswedel

 It is here! KNIME 2.12 has just been released on July 31!

Access NoSQL databases & JSON

With the new MongoDB Integration, a full suite of dedicated nodes has been made available to access MongoDB. As MongoDB uses JSON documents to store data, the JSON category has consequently been expanded to host even more functionalities to process JSON structures: compose from table, JSON to table and vice versa, combine & write, and more.  The JSON data type and the corresponding nodes are the key to integrating tools like MongoDB . They are also used to allow a KNIME workflow being exposed as a RESTful web service via the KNIME server.

JavaScript based nodes & web portal dashboards

Another hot new entry to the tool integration section is a new JavaScript integration. Apart from the “Generic JavaScript” node (which now also has an image output port) KNIME 2.12 comes with a new extension API that enables developers to contribute JavaScript pieces as opposed to plain Java nodes. There is a broad variety of JS libraries available, most prominently D3 and its variants, which can be easily utilized when defining new view nodes. Other than their great appearance and interactivity they have two important benefits: They have finely grained view controls, which are now part of the user configuration, and they run in the browser. These nodes can be easily deployed in the KNIME web portal and be used to build simple dashboards very easily.

New GUI Features

For the lovers of graphical interfaces among us, KNIME Analytics Platform 2.12 has been provided with two more helpful GUI features. A node can be dragged and dropped from the Node Repository to either a connection or an existing node. In the first case, the new node breaks the connection and is inserted between the two originally connected nodes. In the second case, the existing node is removed and replaced by the new one.

Analytics

This new version of KNIME Analytics Platform also likes rules. A lot! First of all, it is now possible to export a decision tree model into a rule set with the Decision Tree to Rule Set node. The output data table contains the decision tree generated rules, in the form <antecedent> => <consequent>, which is the same easy to interpret form used in the Rule Engine node. Secondly, it is also possible to apply this or any previously generated rule set to a new data table using the Rule Engine (Dictionary) node.

A few statistical measures and some new statistics nodes have also been added respectively to the aggregation method menu of the GroupBy node and to the KNIME Labs/Statistics category.

To end with a few more gems: the Missing Values node is now PMML compatible and comes with an Applier node; there are a few more nodes for modular data generation and additional CASE Switch structures.

The KNIME Server as an Orchestration Tool

The new functionality of 2.12 is not limited to KNIME Analytics Platform. KNIME Server and the KNIME Local Automation products also have many new features to offer, from exporting a workflow through a REST interface to the WebPortal branding and the possibility of calling out to other workflows.

In addition, a node called Wait… allows waiting for time to pass or for events to happen, which is a further step towards an event guided architecture.

Streaming Execution

Another interesting feature is the free DataFlow execution engine, introduced by our partner Actian and available for free for KNIME Analytics Platform.  It provides a set of performance optimized ETL  and analytics nodes as well as a streaming executor that increases throughput.

For more details about the new features introduced in KNIME 2.12 please check What is new in KNIME 2.12.

Further Reading: