PMML 4.3 - Changes from PMML 4.2.1
PMML4.3 Menu

Home

Changes

XML Schema

Conformance

Interoperability

General Structure

Field Scope

Header

Data
Dictionary


Mining
Schema


Transformations

Statistics

Taxomony

Targets

Output

Functions

Built-in Functions

Model Verification

Model Explanation

Multiple Models

Association Rules

Baseline Models

Bayesian Network

Cluster
Models


Gaussian
Process


General
Regression


k-Nearest
Neighbors


Naive
Bayes


Neural
Network


Regression

Ruleset

Scorecard

Sequences

Text Models

Time Series

Trees

Vector Machine

PMML 4.3 - Changes from PMML 4.2.1

New Model Types

General Structure

  • Allowed PROB-NUMBER to accept scientific notation.
  • Added GaussianProcess to MODEL-ELEMENT
  • Added BayesianNetwork to MODEL-ELEMENT

Header

  • Added new attribute modelVersion.

Mining Schema

  • Clarified when lowValue and highValue are required.

Built-in Functions

  • Added new functions:
    normalCDF
    Normal Cumulative Density Function
    normalPDF
    Normal Probability Density Function
    stdNormalCDF
    Standard Cumulative Density Function
    stdNormalPDF
    Standards Normal Probability Density Function
    erf
    Related Error Function
    normalIDF
    Normal Inverse Distribution Function
    stdNormalIDF
    Standard Normal Inverse Distribution Function
  • Corrections to documentation.

Output

  • Made dataType of OutputField required.
  • Added a new attribute isFinalResult to OutputField.
  • Added mentions of KNN models with regard to the rank attribute and the entityId feature.

Model Explanation

  • Allowed multiple instances of LiftData in PredictiveModelQuality.

Scope of Fields

  • Allowed an OutputField with feature="transformedValue" to forward reference a derived field defined in LocalTransformations.

Association Rules

  • Added new attributes field and category to Item element.
  • Augmented documentation.

General Regression

  • Added clarifications to the documentation.

KNN

  • Corrected record and field counts in Scoring Example 1.

Multiple Models

  • Clarifications to various model combination methods.
  • Defined how predicted probabilities are to be calculated for the following model combination methods:
    • majorityVote
    • weightedMajorityVote
    • average
    • weightedAverage
    • max
    • median

Neural Networks

  • New activation function rectifier.

Support Vector Machines

  • Added a new attribute maxWins to SupportVectorMachineModel.
  • Added a new element CategoricalPredictor to VectorFields.
  • Dropped an inadvertent requirement to specify classificationMethod on regression models.
  • Added CategoricalPredictor as an alternative to FieldRef in VectorFields element.

Regression

  • Major changes to the documentation regarding classification model scoring.

Rule Sets

  • Minor corrections to the documentation.

Transformations

  • Removed fixed method="indicator" attribute in element NormDiscrete as it served no useful purpose.
  • Added specification for lagged fields.
  • Minor corrections to the documentation.