Today we are announcing new IBM SPSS extensions to connect to Zementis ADAPA services. ADAPA (Adaptive Decision and Predictive Analaytics) uses predictive models to score data. It is based on Predictive Model Markup Language (PMML) standard to import and deploy predictive models. Now with these extensions you can directly upload your IBM SPSS models directly to ADAPAand deploy, execute and manage your predictive models.

adapa

These set of extensions are used as connectors to Zementis ADAPA services:

  • Zementis Model: Score your data with a PMML model previously uploaded.
  • Upload PMML: Create a predictive model on IBM SPSS Modeler and upload it to Zementis.
  • Get Model Info: Get all the input/output variables of your model and its type.

zementisExtensions

The extensions are available on our Github account: http://ibmpredictiveanalytics.github.io/

To follow along with these instructions, you can download the sample data and stream here

  1. Upload PMML: Create a predictive model on IBM SPSS Modeler and upload it to Zementis. This extension can also be used to delete models in the Zementis cloud.

User Interface:

  • Double click on the node to get the options. There are the following fields:
    • Username
    • Password
    • Select Action: There are two options, ‘Upload’ or ‘Delete’.
    • Browse for a PMML File: Select the PMML file to upload
    • Model (only for Deleting): In case ‘Delete’ option is selected, specify here the name of the model to be removed.

When uploading a new model, the name will come from the PMML file. In SPSS this is the name of the model.

Example:

    • Import data from IrisFull.csv using the „Var. File’ source node.
    • Connect  a ‘Type’ node and select ‘setosa’ as Target.

zementisExtensions1

    • Connect  a ‘Type’ node and select ‘setosa’ as Target.

zementisExtensions2

    • Double click on the model node that was generated and click File export PMML.

 

zementisExtensions3

    • Connect the ‘Upload PMML’ node to upload the node to Zementis cloud.

zementisExtensions4

2. Get Model Info: Get all the input/output variables of your model and its type. This node is important to know exactly the input parameters to be able to score using ‘Zementis Model’.

User Interface:

    • Double click on the node to get the options of the node. There are two fields, ‘username’ and ‘password’.

zementisExtensions5

Example:

    • Connect the ‘Get Zementis Models’ to any node in IBM SPSS Modeler.
    • Run the stream. The result will be all the Models available and all their parameters and types.

zementisExtensions6

3. Zementis Model: Score your data with a PMML model previously uploaded.

User Interface:

    • Double click on the node to get to the options. There are 3 fields to be filled: username, password, and Model Name.

zementisExtensions7

Example:

    • Import data from Iris1.csv using the ‘Var. File’ source node.
    • Connect the ‘Zementis Model’ node and put the correct username, password and Model Name. The inputs for the model have to be the same as the inputs of the dataset.
    • The result is going to be the input dataset and some new fields with the prediction and the probability of each of the inputs.

zementisExtensions8

 

Join The Discussion

Your email address will not be published. Required fields are marked *