The Clustering Accelerator deploys the Clustering Model Class and the related logics.

Prerequisites

Before you start the installation of the accelerator, ensure you have the according Product Data Source and potentially a Transactions Datamart or Data Source.

For some important points about it, see Data Requirements (Optimization - Product Similarity).

Deployment

1. Get access to PlatformManager and target partition, as described in common installation prerequisites.

2. Go to PlatformManager at https://platform.pricefx.com/ and log in.

3. Go Marketplace and find the Optimization - Product Similarity accelerator.

4. Click the accelerator tile, select the partition where you want to deploy the accelerator package and confirm the deployment dialog to start.
   For detailed description of all deployment options, see PlatformManager documentation.

5. Set up Datamart mapping.

   • Some rules apply in the mapping:

     • The numerical values must be extended.

     • The percentage values must be values between 0 and 1 (i.e. a margin rate is defined by margin/revenue).

6. Click Continue and wait until the deployment is complete, that is:

   • ProductSimilarity model class is deployed.

   • All 22 required logics and libraries are deployed and added to the partition.

   • PythonEngine is configured to operate on the partition.

Python Engine Setup

The Python Engine, datascience flavor, is required to run some steps of the model. It is deployed and configured automatically with the accelerator. You may just need to wait a few minutes to get it enabled on your partition if that was not the case earlier on.

Also note, it is better to deploy the accelerator to a new partition than copy the logic from one partition to another, because then the engine will not be enabled and configured.

It is possible to update the Python Engine configuration on the partition, if needed. Allowed people can use the functionality in Platform Manager (the image to allow is cregistry.pricefx.eu/engineering/pricefx-python-engine/neural)