Using RTD Load Generator to train learning models

The RTD tool: load generator can be used to train models with historical data to have an converged model before go-live. Otherwise the models will start by using random recommendations until a pattern is discovered.

The load generator is located in the following location:

image

The loadgen.cmd also calls the SetSDParams.cmd file. The file SetSDParams.cmd can be edited to with a line:

SET JAVA_HOME=C:\Program Files (x86)\Java\jdk1.7.0_80

to reference the correct JAVA_HOME location as shown below:

image

Once the load generator starts, the following configurations need to be done:

  1. specify path to load generator xml file (containing the training data)
  2. specify connection details to RTD server

to specify the path click “file open” and select the xml file:

image

Once the xml file is open in the load generator the connection properties need to be set using the http configuration file:

image

The configuration file should look similar to:

image

The load generator script can be started by pressing the play button and the tool will create requests using the supplied xml historical learning data:

image

When now accessing the RTD Decision Center, the models are already trained with the supplied data:

image

e.g. individuals who are best-fit for the product “Credit Protection”.

image