New Tanagra 1.4.23 version

View previous topic View next topic Go down

New Tanagra 1.4.23 version

Post  Stéphane on Sat May 24, 2008 10:10 am

The last Tanagra release (1.4.23, may 2008) introduce new supervised learning components which could be usefull for plankton recognition : PLS discricrimant analysis and PLS linear discriminant analysis. This component have not been tested yet in the context of Plankton Identifier.
Don't hesitate to build your own tdm files to test these new components and to share your experience.
If successfull, new tdm files will be added to next PkID release.

For more information about tanagra 1.4.23 Click here

_________________
Stéphane Gasparini,
Developper of Plankton Identifier
avatar
Stéphane

Posts : 7
Join date : 2008-05-05
Location : LOV, Villefranche-sur-mer, France

View user profile http://www.obs-vlfr.fr/~gaspari/Plankton_Identifier/index.php

Back to top Go down

Tdm for Tanagra 1.4.23

Post  Stéphane on Thu Jun 05, 2008 3:52 pm

Dear users,

for people who want to evaluate performance of PLS-DA and/or PLS-LDA for plankton recognition with their existing test pid, I build the following diagram :

[Diagram]
Title=Plankton Identification REPORT (Test PLS)
Database=

[Dataset]
MLClassGenerator=TMLGenDataset
successors=1
succ_1=Discrete select examples 1

[Discrete select examples 1]
MLClassGenerator=TMLGenCompISAttValue
attribute=Status
value=Learning
successors=1
succ_1=Define status 1

[Define status 1]
MLClassGenerator=TMLGenFSDefStatus
target_count=1
target_1=Ident
illus_count=0
successors=1
succ_1=Supervised Learning 1 (PLS-DA)

[Supervised Learning 1 (PLS-DA)]
MLClassGenerator=TMLGCompOneInstance
embedded_spv=1
embedded_section=Supervised Learning 1 (PLS-DA)--PLS-DA
successors=1
succ_1=Supervised Learning 2 (PLS-LDA)

[Supervised Learning 1 (PLS-DA)--PLS-DA]
MLClassGenerator=TMLGCompSpvPLSDA
nb_axis=0
rdy_cut=0.025

[Supervised Learning 2 (PLS-LDA)]
MLClassGenerator=TMLGCompOneInstance
embedded_spv=1
embedded_section=Supervised Learning 2 (PLS-LDA)--PLS-LDA
successors=2
succ_1=Define status 2
succ_2=Export dataset 1

[Supervised Learning 2 (PLS-LDA)--PLS-LDA]
MLClassGenerator=TMLGCompSpvPLSLDA
nb_axis=0
rdy_cut=0.025

[Define status 2]
MLClassGenerator=TMLGenFSDefStatus
target_count=1
target_1=Ident
input_count=2
input_1=pred_SpvInstance_1
input_2=pred_SpvInstance_2
illus_count=0
successors=1
succ_1=Test 1

[Test 1]
MLClassGenerator=TMLGenCompSpvAssesTestSet
examples=1
successors=0

[Export dataset 1]
MLClassGenerator=TMLGenCompExportData
SelectExamples=0
SelectAttributes=0
Filename=
successors=0


To use it :
- copy/paste the above text (in red) in the notepad
- save it as " test x (PLS).tdm "
- then import it in PkID using tools->import tanagra data mining diagram.

This tdm requires Tanagra 1.4.23 or above (1.4.24 is now available and is still fully compatible with PkID 1.2.6 release 2)
Remember that such test method only apply to pid file gathering objects with the status "learning" for model construction and object with the status "test" for model evaluation (see PkID user guide for details).

feel free to propose your own versions Very Happy

_________________
Stéphane Gasparini,
Developper of Plankton Identifier
avatar
Stéphane

Posts : 7
Join date : 2008-05-05
Location : LOV, Villefranche-sur-mer, France

View user profile http://www.obs-vlfr.fr/~gaspari/Plankton_Identifier/index.php

Back to top Go down

Complete set of tdm using PLS

Post  Stéphane on Thu Jun 05, 2008 6:06 pm

for interrested people, please find a complete set of tdm for PkID using PLS components of tanagra 1.4.23 and above :


Cross-validation 8 (PLS-DA).tdm

[Diagram]
Title=Plankton Identification REPORT (Cross-validation PLS-DA)
Database=

[Dataset]
MLClassGenerator=TMLGenDataset
successors=2
succ_1=Define status 1
succ_2=Define status 2

[Define status 1]
MLClassGenerator=TMLGenFSDefStatus
target_count=1
illus_count=0
successors=1
succ_1=Supervised Learning 1 (PLS-DA)

[Supervised Learning 1 (PLS-DA)]
MLClassGenerator=TMLGCompOneInstance
embedded_spv=1
embedded_section=Supervised Learning 1 (PLS-DA)--PLS-DA
successors=1
succ_1=Cross-validation 1

[Supervised Learning 1 (PLS-DA)--PLS-DA]
MLClassGenerator=TMLGCompSpvPLSDA
nb_axis=0
rdy_cut=0.025

[Cross-validation 1]
MLClassGenerator=TMLGenCompAssesCV
isSaveResults=0
nb_repetitions=5
nb_folds=2
successors=0

[Define status 2]
MLClassGenerator=TMLGenFSDefStatus
target_count=0
input_count=1
input_1=Ident
illus_count=0
successors=1
succ_1=Univariate discrete stat 1

[Univariate discrete stat 1]
MLClassGenerator=TMLGenCompSDUnivDisc
sort_result=0
sort_criteria=0
successors=1
succ_1=Export dataset 1

[Export dataset 1]
MLClassGenerator=TMLGenCompExportData
SelectExamples=0
SelectAttributes=0
Filename=
successors=0



Cross-validation 9 (PLS-LDA).tdm

[Diagram]
Title=Plankton Identification REPORT (Cross-validation PLS-LDA)
Database=

[Dataset]
MLClassGenerator=TMLGenDataset
successors=2
succ_1=Define status 1
succ_2=Define status 2

[Define status 1]
MLClassGenerator=TMLGenFSDefStatus
target_count=1
illus_count=0
successors=1
succ_1=Supervised Learning 1 (PLS-LDA)

[Supervised Learning 1 (PLS-LDA)]
MLClassGenerator=TMLGCompOneInstance
embedded_spv=1
embedded_section=Supervised Learning 1 (PLS-LDA)--PLS-LDA
successors=1
succ_1=Cross-validation 1

[Supervised Learning 1 (PLS-LDA)--PLS-LDA]
MLClassGenerator=TMLGCompSpvPLSLDA
nb_axis=0
rdy_cut=0.025

[Cross-validation 1]
MLClassGenerator=TMLGenCompAssesCV
isSaveResults=0
nb_repetitions=5
nb_folds=2
successors=0

[Define status 2]
MLClassGenerator=TMLGenFSDefStatus
target_count=0
input_count=1
input_1=Ident
illus_count=0
successors=1
succ_1=Univariate discrete stat 1

[Univariate discrete stat 1]
MLClassGenerator=TMLGenCompSDUnivDisc
sort_result=0
sort_criteria=0
successors=1
succ_1=Export dataset 1

[Export dataset 1]
MLClassGenerator=TMLGenCompExportData
SelectExamples=0
SelectAttributes=0
Filename=
successors=0



Spv learning 8 (PLS-DA).tdm

[Diagram]
Title=Plankton Identification REPORT (spv learning PLS DA)
Database=

[Dataset]
MLClassGenerator=TMLGenDataset
successors=1
succ_1=Discrete select examples 1

[Discrete select examples 1]
MLClassGenerator=TMLGenCompISAttValue
attribute=Status
value=Learning
successors=1
succ_1=Define status 1

[Define status 1]
MLClassGenerator=TMLGenFSDefStatus
target_count=1
target_1=Ident
illus_count=0
successors=1
succ_1=Supervised Learning 1 (PLS-DA)

[Supervised Learning 1 (PLS-DA)]
MLClassGenerator=TMLGCompOneInstance
embedded_spv=1
embedded_section=Supervised Learning 1 (PLS-DA)--PLS-DA
successors=1
succ_1=Recover examples 1

[Supervised Learning 1 (PLS-DA)--PLS-DA]
MLClassGenerator=TMLGCompSpvPLSDA
nb_axis=0
rdy_cut=0.025

[Recover examples 1]
MLClassGenerator=TMLGenCompISRecoverExamples
recov_prm=1
successors=2
succ_1=Define status 2
succ_2=Export dataset 1

[Define status 2]
MLClassGenerator=TMLGenFSDefStatus
target_count=0
input_count=1
input_1=pred_SpvInstance_1
illus_count=0
successors=1
succ_1=Univariate discrete stat 1

[Univariate discrete stat 1]
MLClassGenerator=TMLGenCompSDUnivDisc
sort_result=1
sort_criteria=1
successors=0

[Export dataset 1]
MLClassGenerator=TMLGenCompExportData
SelectExamples=1
SelectAttributes=0
Filename=
successors=0



Spv learning 9 (PLS-LDA).tdm

[Diagram]
Title=Plankton Identification REPORT (spv learning PLS LDA)
Database=

[Dataset]
MLClassGenerator=TMLGenDataset
successors=1
succ_1=Discrete select examples 1

[Discrete select examples 1]
MLClassGenerator=TMLGenCompISAttValue
attribute=Status
value=Learning
successors=1
succ_1=Define status 1

[Define status 1]
MLClassGenerator=TMLGenFSDefStatus
target_count=1
target_1=Ident
illus_count=0
successors=1
succ_1=Supervised Learning 1 (PLS-LDA)

[Supervised Learning 1 (PLS-LDA)]
MLClassGenerator=TMLGCompOneInstance
embedded_spv=1
embedded_section=Supervised Learning 1 (PLS-LDA)--PLS-LDA
successors=1
succ_1=Recover examples 1

[Supervised Learning 1 (PLS-LDA)--PLS-LDA]
MLClassGenerator=TMLGCompSpvPLSLDA
nb_axis=0
rdy_cut=0.025

[Recover examples 1]
MLClassGenerator=TMLGenCompISRecoverExamples
recov_prm=1
successors=2
succ_1=Define status 2
succ_2=Export dataset 1

[Define status 2]
MLClassGenerator=TMLGenFSDefStatus
target_count=0
input_count=1
input_1=pred_SpvInstance_1
illus_count=0
successors=1
succ_1=Univariate discrete stat 1

[Univariate discrete stat 1]
MLClassGenerator=TMLGenCompSDUnivDisc
sort_result=1
sort_criteria=1
successors=0

[Export dataset 1]
MLClassGenerator=TMLGenCompExportData
SelectExamples=1
SelectAttributes=0
Filename=
successors=0



enjoy ! Wink

_________________
Stéphane Gasparini,
Developper of Plankton Identifier
avatar
Stéphane

Posts : 7
Join date : 2008-05-05
Location : LOV, Villefranche-sur-mer, France

View user profile http://www.obs-vlfr.fr/~gaspari/Plankton_Identifier/index.php

Back to top Go down

Re: New Tanagra 1.4.23 version

Post  Sponsored content


Sponsored content


Back to top Go down

View previous topic View next topic Back to top

- Similar topics

 
Permissions in this forum:
You cannot reply to topics in this forum