Summer Update of Extensions

MartinLiebig

Shortly before our release of RapidMiner 9.7 we are also proud to announce the most recent update on our extensions Operator Toolbox and Converters.

Operator Toolbox is our staging area for new and helpful operators. With the release of version 2.5. we enhance some of the existing operators to make them more useful and add support to read Word-Files. In Converters extension we add a conversion operator to use Automatic Feature Engineering in conjunction with Model Simulators.

Read Word Files with Read Doc(x)

 

The new operator Read Doc(x) allows you to read word files. It simply allows you to read Documents created by MS Word into RapidMiner. The operator can use Fileobjects as input, so you can easily use Loop Files to loop over a folder of documents and read them in.

Feature Set and Model Simulators

Automatic Feature Engineering is a powerful way to make your models better. Previously the Model Simulator for Models which used Feature Engineering were a bit hard to use. They show the transformed new features and not the original features.

The new operator “FeatureSet to Model” allows you to convert the FeatureSet from Automatic Feature Engineering to a Preprocessing Model. You may already know Preprocessing Models from operators like Normalize. You can group those models together with your prediction model, to get one chained model. 

The result is a model simulator which uses the results of Automatic Feature Engineering, but it shows the original values to work with:

Build Simulation with Constant Attributes

In our recent 2.4 update we added the new Operator Build Simulation to Operator Toolbox. It allows you to generate new examples, which have the same statistical properties as the examples you trained it on. This can be very handy if you need to generate pseudo examples for data privacy reasons, but also if you want to see the effect of certain parameter ranges on your model.

In the new update we added the option to make certain attributes constant. This allows you for example to generate data only for 3rd class passengers of Titanic, while still using the full data set for training your simulation model.

Better SFTP Support

There are two new operators to work with SFTP Servers. Read SFTP and Write SFTP replace the old SFTP Download File and SFTP Upload File. The new operators can work with the new connection management framework introduced in version 9.3.

Both operators are now supporting HTTP proxies. We also changed the file handling of the operator, so that you get file objects back which you can use in a more versatile way.

 

Get More Information on your Association Rules

Association Rules are usually used in Market Basket analysis. If somebody bought Coke and Chips, (s)he may also buy Pizza. If you want to learn more about Market Basket Analysis, check out our Lightning Demo on July 7th!

The Apply Association rule operator is telling you to offer that person Pizza. With the new Apply Association Rules (Detailed) you get more information on why this recommendation is done. You get all applicable rules and their respective statistics.