?????????????PMML(Predict Model Markup Language)

1
?????????????PMML(Predict Model Markup Language)

• ??PMML
• ????
• ???????
• PMML???
• PMML DTD

2
??PMML

• ????XML????????
• ????????(Data Mining)???
• ????XML???,??????XML??????
• ??????????
• ??????????
• ???????????

3
??PMML(?)

• ??????????(Nation Center for Data Mining)??????
• ????www.dmg.org
• ???????????????
• DMG PMML
• OMG CWM DM
• SQL/MM Part 6 for Data Mining
• JSR-073 Java Data Mining API
• Microsoft OLE DB for Data Mining

4
Data Mining Group

• PMML Version 1.1????
• Angoss, IBM, Magnify, Microsoft, NCR, Oracle,
  SPSS, University of Illinois at Chicago
• Focused group to expedite process
• PMML Version 1.2 ??Xchange,?????
• Open to any qualified vendor selling data mining
  products
• Augmented by experts reviewers
• ??xml.org?????????

5
PMML???

• PMML??????????????????
• PMML 1.0??????????(Model),????????
• ?????????????
• PMML??????MetaData??????
• ???????????????
• ??????????

6
PMML???(?)

• Open standard for Data Mining Models
• Not is concerned with the process of creating a
  model
• Provides independence from application, platform,
  and operating system
• Simplifies use of data mining models by other
  applications (consumers of data mining models)

7
????

• ???????(Data Mining)????????????,????????Data
  Mining????????????????????????????????????????????
  ????,??????????????????????????,??????????????????
  ???????????,?Data Mining???????,?????????,??????,?
  ?Internet??????????,????????????

8
????(?)

• ??Data Mining?????????????????????????????????????
  ????????,????????????????????????????????????????
  ??????????????,???????????????????????????,???????
  ????,???????????????????,?????????????????????????
  ?,????????,?????????????,????????,????????????

9
???????

• ???(classification)
• ???(estimation)
• ???(prediction)
• ?????(affinity grouping)
• ?????(clustering)

10
???????

• ??????(Association Rule)
• ??????(Automatic Cluster Detection)
• ???(Decision Tree)
• ?????(artificial neural network)
• ???????(Genetic Algorithm)
• ??????(Online Analytical Processing)

11
PMML???(v1.1)

• Schemas
• Data dictionary (data schema, including outliers,
  missing values)
• Mining schema
• Infrastructure
• Univariate statistics
• Normalization and transformation (very basic)
• Models

12
PMML???(v1.1)

• Polynomial regression
• General regression
• Trees
• Center based clusters
• Density based clusters
• Associations
• Neural nets
• more to be added in v1.2

13
?? PMML (v1.1)

• ltTreeModel modelName"golfing"gt ltMiningSchemagt
  ltMiningField name"temperature"/gtltMiningField
  name"humidity"/gt ...lt/MiningSchemagt ltNode
  score"play"gt ltPredicate field"outlook"
  operator"equal" value"sunny"/gtltNode
  score"play"gtltCompoundPredicate
  booleanOperator"and" gtltPredicate
  field"temperature operator"lessThan"
  value"90F" /gt ltPredicate field"temperature"
  operator"greaterThan" value"50F" /gt

14
PMML DTD_v1_1

• lt?xml version'1.0' encoding'ISO-8859-1'
  ?gtlt!ENTITY A-PMML-MODEL '(TreeModel
  NeuralNetwork ClusteringModel
  RegressionModel GeneralRegressionModel
  AssociationModel )' gtlt!ELEMENT PMML ( Header,
  DataDictionary, (A-PMML-MODEL), Extension )
  gtlt!ATTLIST PMML version CDATA
  REQUIREDgtlt!ELEMENT Extension ANY gtlt!ATTLIST
  Extension extender CDATA IMPLIEDname CDATA
  IMPLIED value CDATA IMPLIED gt

15
PMML DTD_v1_1(?)

• lt!ENTITY NUMBER "CDATA" gtlt!ENTITY INT-NUMBER
  "CDATA"gtlt!-- content must be an integer, no
  fractions or exponent --gtlt!ENTITY REAL-NUMBER
  "CDATA"gtlt!-- content can be any number covers
  C/C types 'float','long','double'scientific
  notation, eg 1.23e4, is allowed --gtlt!ENTITY
  PROB-NUMBER "CDATA"gtlt!-- a REAL-NUMBER between
  0.0 and 1.0usually describing a probability
  --gtlt!ENTITY PERCENTAGE-NUMBER "CDATA"gtlt!-- a
  REAL-NUMBER between 0.0 and 100.0 --gt

16
PMML DTD_v1_1(?)

• lt!ENTITY FIELD-NAME "CDATA"gtlt!ELEMENT Array
  (PCDATA) gtlt!ATTLIST Arrayn INT-NUMBER
  IMPLIEDtype ( int real string )
  IMPLIEDgtlt!ENTITY NUM-ARRAY "Array"gtlt!-- an
  array of numbers --gtlt!ENTITY INT-ARRAY
  "Array"gtlt!-- an array of integers --gtlt!ENTITY
  REAL-ARRAY "Array"gtlt!-- an array of reals
  --gtlt!ENTITY STRING-ARRAY "Array"gtlt!-- an
  array of strings --gt