Analytics Scoring Components

Analytics Scoring consists of the following components.

• Binning Lookup—This stage can be used to apply previously defined binning to new data using existing bins created in dataflows using the Machine Learning Binning stage.
• Java Model Scoring—This stage can be used score new data using the formula created when you fit a machine learning model.
• PMML Model Scoring—This stage can be used to evaluate any model stored in the Analytics Scoring Repository in the context of a dataflow.
• Read from Miner Dataset—This stage can be used to read data from a focus file to be used within a dataflow.
• Write to Miner Dataset—This stage can be used to write data from a dataflow to a focus file.
• Machine Learning Model Management—This repository includes Model Assessment, where you manage all machine learning models on your Spectrum Technology Platform server, and Binning Management, where you manage all binning on your Spectrum Technology Platform server.
• Analytics Scoring Repository —This is the central repository for all models available to Analytics Scoring. Users can manage the repository via a web client.

See "Data Science Demonstration Flows" in the Machine Learning Guide for examples of supervised and unsupervised learning that include the scoring of data using Java Model Scoring.