New Data Miner Features in SQL Developer 4.0

These new Data Miner 4.0 features are supported for all database versions supported by Data Miner (Oracle Database 11.2.0.1 and later):

• Workflow SQL Script Deployment

  Generates SQL scripts to support full deployment of workflow contents

• SQL Query Node

  • Integrate SQL queries to transform data or provide a new data source

  • Supports the running of R Language Scripts and viewing of R generated data and graphics

• Graph Node

  Generate Line, Scatter, Bar, Histogram and Box Plots

• Workflow Performance Features

  • Parallel Processing: Specify degree of parallel query processing desired per workflow node.

  • Table Compression: Table generation utilizes table compression feature; see Create Table and Compression.

• Model Build Node Improvements

  • Node level data usage specification applied to underlying models

  • Node level text specifications to govern text transformations

  • Displays heuristic rules responsible for excluding predictor columns

  • Ability to control the amount of Classification and Regression test results generated

  • Ability to turn the generation of Tuning results on and off

• View Data

  Ability to drill in to view custom objects and nested tables

• Model Details Node

  • Added Cluster Centroid details

• Explore and Transform Nodes

  • Improvements in NULL handling

• Explore Node

  • Ability to select specific statistical outputs

• Workflow Run and Validation Options

  • Ability to select multiple nodes when invoking Workflow Run options

  • Added Validation options allowing users to run the Workflow in validation mode (Validate Parents)

• Workflow Import

  • Validate if the target database version supports the functionality contained in the imported workflow

• Model Tuning

  • Added option to control the generation of results used for post-model build tuning

• Column Filter Node

  • Added System Determined option for Attribute Importance sampling technique

• Model Test Performance

  • Specifying Case ID improves performance of Classification and regression testing

• Viewers and Editors

  • Removed blocking dialogs triggered by long-running processes and queries

• Copy and Paste improvements

  • Ability to copy charts to the Windows Clipboard or to a file

  • Ability to copy data grids to the Windows Clipboard

  • Ability to easily copy Cluster and Decision Tree Rules to the Windows Clipboard or to a file

  • Ability to copy and paste workflows between different Data Miner repositories as long as the target repository is compatible with the source repository

• Charts

  • Ability to view and copy data content of all Charts.

• Demo Data

  • Added ODMR_CARS_DEMO table to the demo data scripts

• Backup and Recovery

  • New scripts to provide workflow-level backup and recovery options described in Backup and Restore Oracle Data Miner Repository

This feature requires connection to Oracle Database 11g Release 2 (11.2.0.4) or higher

• Significant improvements to Data Miner Repository performance for workflow save, run, and load times

These features require connection to Oracle Database 12c or later:

• Predictive Query Nodes

  • Predictive results without the need to build models using Analytical Queries

  • Refined predictions based on data partitions

• Clustering Node New Algorithm

  AddedExpectation Maximization algorithm

• Feature Extraction Node New Algorithms

  Added Singular Value Decomposition and Principal Components Analysis algorithms

• Text Mining Enhancements

  • Text transformations integrated as part of Model's Automatic Data Preparation

  • Ability to import Build Text node specifications into a Model Build node

• Prediction Result Explanations

  Scoring details that explain predictive result

• Generalized Linear Models New Algorithm Settings

  New algorithm settings provide feature selection and generation

• Extended Data Type Support

  Support for VARCHAR2 size of 32767.