workflow.xml

<tool id='sl_core' name='sl_core'>
  <command interpreter='bash'>workflow.sh</command>
  <inputs>
    <section name='repos' type='section' title='Repositories, branches, resources, and environments' expanded='true'>
      <param name='ml_arch_repo'
	     label='ML archive repository'
	     type='text'
	     value='git@github.com:parallelworks/dynamic-learning-rivers'
	     width='50%_none%'
	     help='Repo MUST have SSH prefix, e.g. git@github.com:org/repo, for SSH auth w/ deploy keys to push'>
      </param>
      <param name='ml_arch_branch'
	     label='ML archive BRANCH'
	     type='text'
	     value='test-auto-workflow'
	     width='50%_node%'>
      </param>
      <param name='push_to_gh'
             type='boolean'
             truevalue='Yes'
             falsevalue='No'
             checked='True'
             label='Push results to GitHub'
             help='Select yes to push results to GitHub. A deploy key for the repo is required.'
             width='25%_none%'
             float='right'>
      </param>
      <param name='ml_code_repo'
	     label='ML code repository'
	     type='text'
	     value='https://github.com/parallelworks/sl_core'
	     width='50%_none%'>
      </param>
      <param name='ml_data_repo'
	     label='ML data repository'
	     type='text'
	     value='https://github.com/parallelworks/global-river-databases'
	     width='50%_none%'>
      </param>
      <param name='whost'
	     label='Workflow host (cluster name)'
	     type='computeResource'
	     hideDisconnectedResources='False'
	     width='50%_none%'
	     help='Name of cluster to run on. Select a resource from drop down menu.'>
      </param>
      <param name='miniconda_loc'
	     label='Location (prefix) of Miniconda installation'
	     type='text'
	     value='/home/__USER__/.miniconda3'
	     width='50%_none%'
	     help='Location of miniconda on cluster.  __USER__ is auto substituted.'>
      </param>
      <param name='my_env'
	     label='Name of Miniconda environment'
	     type='text'
	     value='superlearner'
	     width='50%_none%'
	     help='Specify name of Miniconda environment to use. There is a bug with using base!'>
      </param>
    </section>
    <section name='superlearner' type='section' title='SuperLearner configuration' expanded='False'>
      <param name='train_test_data'
	     label='Data for training and testing [.csv] [Inputs, Outputs]'
	     type='text'
	     help='The header contains the parameter names; the inputs must occupy the first columns; path wrt ML archive repo.'
	     value='scripts/prep_06_output_final_train.csv'
	     width='50%'>
      </param>
      <param name='predict_data'
             label='Data for predicting [.csv] [Inputs, Outputs]'
             type='text'
             help='The header contains the parameter names; only inputs; path wrt ML archive repo.'
             value='scripts/prep_06_output_final_predict'
             width='50%'>
      </param>
      <param name='num_inputs'
	     label='Number of inputs'
	     type='text'
	     value='25'
	     width='25%'>
      </param>
      <param name='superlearner_conf'
	     label='SuperLearner configuration'
	     type='text'
	     help='SuperLearner configuration; path wrt ML code repo'
	     value='sample_inputs/superlearner_conf.py'
	     width='50%_none%'>
      </param>
      <param name='work_dir_base'
             label='Working directory basename wrt archive repo'
             type='text'
             value='ml_models/sl_'
             width='50%_none%'
	     help='Prefix/base of the location where to store all SL output in the archive repo. Test!'>
      </param>
      <param name='hpo'
	     type='boolean'
	     truevalue='Yes'
	     falsevalue='No'
	     checked='False'
	     label='Run hyperparameter optimization?'
	     help='Select yes for hyperparameter tuning'
	     width='25%_none%'
	     float='right'>
      </param>
      <param name='cross_val_score'
	     type='boolean'
	     truevalue='Yes'
	     falsevalue='No'
	     checked='False'
	     label='Run cross-validation?'
	     help='Select yes to run cross-validation on the entire data set'
	     width='25%_none%'
	     float='right'>
      </param>
      <param name='smogn'
             type='boolean'
             truevalue='Yes'
             falsevalue='No'
             checked='False'
             label='Use SMOGN to build synthetic data'
             help='Select yes to augment training data with synthetic data.'
             width='25%_none%'
             float='right'>
      </param>
      <param name='onnx'
             type='boolean'
             truevalue='Yes'
             falsevalue='No'
             checked='False'
             label='Save SuperLearner models in ONNX format'
             help='Select yes to save in ONNX format (more portable than .pkl)'
             width='25%_none%'
             float='right'>
      </param>
      <param name='backend'
	     type='select'
	     label='Joblib backend'
	     width='50%'
	     help='Select Backend'
	     multiple='false'>
	<option value='loky' selected='true'>loky</option>
	<option value='multiprocessing'>multiprocessing</option>
	<option value='threading'>threading</option>
	<option value='dask'>dask</option>
      </param>
      <param name='n_jobs'
	     label='Number of parallel jobs on worker'
	     type='integer'
	     min='1'
	     max='100'
	     value='8'
	     width='25%'>
      </param>
      <param name='num_inst'
             label='Number of SuperLearner instances (duplicate stacked-regressors trained)'
             type='integer'
             min='1'
             max='100'
             value='10'
             width='25%'>
      </param>
      <param name='fpi_corr_cutoff'
             label='Correlation cutoff for FPI'
             type='integer'
             min='1'
             max='99'
             value='50'
             width='25%'
             help='Features correlated above this level will be grouped in FPI analysis. This is correlation on scale of 0 to 100 (instead of 0 to 1), but not R^2!'>
      </param>
    </section>
  </inputs>
  <outputs>
  </outputs>
</tool>