|
Job Definition Examples
In this page you will find several examples of job definitions that you can copy and use in your projects. If you would like to add an example to this page or would like us to add a missing example, please contact us.
Basic Jobs
Process a run for 50 generations starting from scratch.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='start' />
</job>
Process a run for 100 generations starting from the current model.
<job filename='MyRunWithModels.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='continue' />
</job>
Simplify the active model for 100 generations.
<job filename='MyRunWithModels.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='simplify' />
</job>
Increase the complexity of the active model for 100 generations.
<job filename='MyRunWithModels.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='complexify' />
</job>
Process 4 runs in parallel from scratch and create a run with the best models (selected by Training Fitness) at the end.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'
usesubfolder='1'
createconsolidatedrun='4'
async='yes'>
<run id='1' stopcondition='generations' value='50' type='start' />
<run id='2' stopcondition='generations' value='50' type='start' />
<run id='3' stopcondition='generations' value='50' type='start' />
<run id='4' stopcondition='generations' value='50' type='start' />
</job>
For other model selection options see the createconsolidatedrun possible values or the job attributes.
Time Series Jobs
Generate a prediction after running for 50 generations and save it to the file prediction.txt
<job filename='MyTimeSeriesRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='start' >
<predict quantity="1" format="text" filename="prediction.txt" />
</run>
</job>
For saving predictions to other formats see here.
Generate a prediction for the model with the best validation fitness after running for 50 generations and save it to the file prediction.txt
<job filename='MyTimeSeriesRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='start' >
<select criteria="bestvalidationfitness"/>
<predict quantity="1" format="text" filename="prediction.txt" />
</run>
</job>
Change the time series mode to prediction (from testing).
<job filename='MyTimeSeriesRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' type='idle'>
<transform name='timeseriestopredictionmode'/>
</run>
</job>
Change the time series mode to testing mode (from prediction) and set the number of testing predictions.
<job filename='MyTimeSeriesRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' type='idle'>
<transform name='timeseriestotestingmode'/>
<settings>
<setting key='TimeSeriesTestingPredictions' value='5'/>
</settings>
</run>
</job>
Loading Data
Load the training set from a database and process the run for 50 generations.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type="start">
<datasets>
<dataset type='training' records='all'>
<connection type="database" format="responselast">
<oledbconnectionstring>{connection string}</oledbconnectionstring>
<sqlstatement>SELECT * FROM TrainSet;</sqlstatement>
</connection>
</dataset>
</datasets>
</run>
</job>
Load the training set from the Excel file file.xlsx and process the run for 50 generations.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50'>
<datasets>
<dataset type='training' records='all'>
<connection type='excel' format='responselast'>
<sheet name='Training'
range='$A$1:$P$11'
columns='all'>C:\examples\file.xlsx</sheet>
</connection>
</dataset>
</datasets>
</run>
</job>
Load the validation set from a GeneXproTools file and process the run for 50 generations.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50'>
<datasets>
<dataset type='training' records='all'>
<connection type="gepfile" gepfiledataset="validation">
<path>C:\examples\file.txt</path>
</connection>
</dataset>
</datasets>
</run>
</job>
Load the training set from a tab separated text file, loading the last 200 records from last to first, and process the run for 50 generations.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50'>
<datasets>
<dataset type='training' records='200' reverse='yes'>
<connection type='file' format='responselast'>
<path separator='tab'
haslabels='yes'>C:\examples\file.txt</path>
</connection>
</dataset>
</datasets>
</run>
</job>
Load the training set directly from this job definition file and process the run for 50 generations.
<job filename='MyRun.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50'>
<datasets>
<dataset type='training' records='all'>
<connection type='internal' format='responselast'>
<path separator='pipe'
haslabels='yes'>Var0|Var1|Response
b|22.22|0
a|58.67|1
a|47.56|0
...
</path>
</connection>
</dataset>
</datasets>
</run>
</job>
Model Selection
Select the model with the best validation fitness and continue the run (improving this model).
<job filename='MyRunWithModels.gep'
path='C:\examples'
feedback='2'>
<run id='1' stopcondition='generations' value='50' type='continue'>
<preselect criteria="bestvalidationfitness"/>
</run>
</job>
Note that all the models must have been tested in a previous occasion for this selection to succeed.
|
