Actian RushAccelerator for KNIME: Scaling KNIME to Big Data

KNIME, the #1 customer satisfaction ranked* open source data mining platform, with thousands of data prep and analytics functions, is now tightly integrated with Actian DataRush, the world’s fastest compute engine for commodity hardware and big data clusters, to provide a depth and breadth of functionality for big data at a price/performance level never seen before. 

Thousands of data analysts use KNIME for every type of analytics, from the most basic scoring to the most sophisticated machine learning and complex object analytics. Actian now brings the high performance and scalability of DataRush to that breadth of capability. Workflows created in KNIME with DataRush enabled flowable nodes can now process data 2-10 times faster on the same hardware. Data sets that used to be prohibitively large become accessible with Actian RushAccelerator for KNIME and data sets that used to require sampling can now be fully analyzed.

KNIME already incorporates over 1000 processing nodes for data I/O, pre-processing and cleansing, modeling, and analysis as well as various interactive views, such as scatter plots, parallel coordinates and others. KNIME integrates all analysis modules of the Weka data mining environment and runs R-scripts, offering access to a vast library of statistical routines.

Now, with Actian RushAccelerator for KNIME, data analysts can give their KNIME workflows a boost of parallel dataflow-enabled speed. 

*Rexer Analytics Data Mining Survey 2011

Download Free Trial of Actian RushAccelerator for KNIME

The analytics library in KNIME supports data mining for marketing, surveillance, fraud detection, cyber security and has an especially comprehensive list of operators for scientific discovery and analysis of pharmaceutical data. Build a recommender system, market basket analysis, customer churn analysis and more using the KNIME extensive library of over 1000 operators. Then, with a small modification to make them flowable, accelerate those operators with Actian RushAccelerator for KNIME

  • Association Rule Mining, Market Basket Analysis, Affinity Analysis
  • Classifiers: Decision tree, K-nearest-neighbors, Naïve Bayes, Support Vector Machine (SVM)
  • Clustering: Recommender learner and predictors, k-means
  • Feature Selection: Principal Component Analysis (PCA)
  • Regression Analysis

Download Free Trial of Actian RushAccelerator for KNIME

Actian RushAccelerator for KNIME is built on the patented Actian DataRush data processing engine to ensure scalability. Actian DataRush automatically detects and utilizes all cores and nodes available up to a settable limit on any machine. Execution moves seamlessly from desktop to server, without the need to modify code, re-design models, or recompile.

At runtime, Actian DataRush automatically parallelizes the work across the available cores. This ensures the most efficient execution. For example, the performance of an application written on a 4-core desktop will automatically almost linearly scale to take full advantage of the additional resources when installed on a 16-core server. Every bit of current hardware will be used to it's fullest capacity, nothing wasted, and applications are future-proofed. Organizations can simply add more compute resources to keep up with growing data volumes.

Download Free Trial of Actian RushAccelerator for KNIME

Need to go beyond the standard set of KNIME operators? 

Actian RushAnalytics for KNIME includes a large set of highly parallel dataflow-enabled operators pre-made. The highly optimized operators can provide an order of magnitude higher speed boost to your KNIME flows. 

If that still isn't enough, you can add new operators to the Actian DataRush library by writing code which calls the Actian DataRush API.  Written in Java, the API is callable from any JVM language, including Java, Javascript, JRuby, Python, Scala and more.

Read more about Actian RushAnalytics for KNIME 


Data Preparation Example [enlarge]

 


Recommender System [enlarge]