annotate search_model_validation.xml @ 1:f727f5ff7d60 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit cfc9fe24b7975fc5838bb3e456646202898eb977

<tool id="sklearn_searchcv" name="Hyperparameter Search" version="@VERSION@">
    <description>using exhausitive or randomized search</description>
    <macros>
        <import>main_macros.xml</import>
    </macros>
    <expand macro="python_requirements">
        <requirement type="package" version="0.9.12">asteval</requirement>
    </expand>
    <expand macro="macro_stdio"/>
    <version_command>echo "@VERSION@"</version_command>
    <command>
        <![CDATA[
        python "$sklearn_search_model_validation_script" '$inputs'
        ]]>
    </command>
    <configfiles>
        <inputs name="inputs" />
        <configfile name="sklearn_search_model_validation_script">
            <![CDATA[
import sys
import json
import pandas
import pickle
import numpy as np
import xgboost
import scipy
from asteval import Interpreter, make_symbol_table
from sklearn import metrics, preprocessing, model_selection, ensemble
from sklearn.pipeline import Pipeline

@COLUMNS_FUNCTION@
@GET_ESTIMATOR_FUNCTION@
@GET_SEARCH_PARAMS_FUNCTION@

input_json_path = sys.argv[1]
with open(input_json_path, "r") as param_handler:
    params = json.load(param_handler)

#handle cheatah
infile1 = "$input_options.infile1"
infile2 = "$input_options.infile2"
infile_pipeline = "$search_schemes.infile_pipeline"
outfile_result = "$outfile_result"
outfile_estimator = "$outfile_estimator"
#if $search_schemes.selected_search_scheme == "RandomizedSearchCV":
np.random.seed($search_schemes.random_seed)
#end if

params_builder = params['search_schemes']['search_params_builder']

input_type = params["input_options"]["selected_input"]
if input_type=="tabular":
    header = 'infer' if params["input_options"]["header1"] else None
    column_option = params["input_options"]["column_selector_options_1"]["selected_column_selector_option"]
    if column_option in ["by_index_number", "all_but_by_index_number", "by_header_name", "all_but_by_header_name"]:
        c = params["input_options"]["column_selector_options_1"]["col1"]
    else:
        c = None
    X = read_columns(
            infile1,
            c = c,
            c_option = column_option,
            sep='\t',
            header=header,
            parse_dates=True
    )
else:
    X = mmread(open("$input_options.infile1", 'r'))

header = 'infer' if params["input_options"]["header2"] else None
column_option = params["input_options"]["column_selector_options_2"]["selected_column_selector_option2"]
if column_option in ["by_index_number", "all_but_by_index_number", "by_header_name", "all_but_by_header_name"]:
    c = params["input_options"]["column_selector_options_2"]["col2"]
else:
    c = None
y = read_columns(
        infile2,
        c = c,
        c_option = column_option,
        sep='\t',
        header=header,
        parse_dates=True
)
y=y.ravel()

optimizers = params["search_schemes"]["selected_search_scheme"]
optimizers = getattr(model_selection, optimizers)

options = params["search_schemes"]["options"]
if 'scoring' in options and options['scoring'] == '':
    options['scoring'] = None
if 'pre_dispatch' in options and options['pre_dispatch'] == '':
    options['pre_dispatch'] = None

with open(infile_pipeline, 'rb') as pipeline_handler:
    pipeline = pickle.load(pipeline_handler)
search_params = get_search_params(params_builder)
searcher = optimizers(pipeline, search_params, **options)

searcher.fit(X, y)

cv_result = pandas.DataFrame(searcher.cv_results_)
cv_result.to_csv(path_or_buf=outfile_result, sep='\t', header=True, index=False)

#if $save:
with open(outfile_estimator, "wb") as output_handler:
    pickle.dump(searcher.best_estimator_, output_handler, pickle.HIGHEST_PROTOCOL)
#end if

            ]]>
        </configfile>
    </configfiles>
    <inputs>
        <conditional name="search_schemes">
            <param name="selected_search_scheme" type="select" label="Select a model selection search scheme:">
                <option value="GridSearchCV" selected="true">GridSearchCV - Exhaustive search over specified parameter values for an estimator </option>
                <option value="RandomizedSearchCV">RandomizedSearchCV - Randomized search on hyper parameters for an estimator</option>
            </param>
            <when value="GridSearchCV">
                <expand macro="search_cv_estimator"/>
                <section name="options" title="Advanced Options for SearchCV" expanded="false">
                    <expand macro="search_cv_options"/>
                </section>
            </when>
            <when value="RandomizedSearchCV">
                <param name="random_seed" type="integer" value="65535" min="0" max="65535" label="Set up random seed:"/>
                <expand macro="search_cv_estimator"/>
                <section name="options" title="Advanced Options for SearchCV" expanded="false">
                    <expand macro="search_cv_options"/>
                    <param argument="n_iter" type="integer" value="10" label="Number of parameter settings that are sampled"/>
                    <expand macro="random_state"/>
                </section>
            </when>
        </conditional>
        <param name="save" type="boolean" truevalue="booltrue" falsevalue="boolflase" checked="true" label="Save the best estimator/pipeline?"/>
        <expand macro="sl_mixed_input"/>
    </inputs>
    <outputs>
        <data format="tabular" name="outfile_result"/>
        <data format="zip" name="outfile_estimator">
            <filter>save</filter>
        </data>
    </outputs>
    <tests>
        <test>
            <param name="selected_search_scheme" value="GridSearchCV"/>
            <param name="infile_pipeline" value="pipeline01"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="C: [1, 10, 100, 1000]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="k: [3, 5, 7, 9]"/>
                <param name="selected_param_type" value="prep_2_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result" >
                <assert_contents>
                    <has_text_matching expression="[^/d]+0.7938837807353147[^/d]+{u'estimator__C': 1, u'preprocessing_2__k': 9}[^/d]+1" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="selected_search_scheme" value="RandomizedSearchCV"/>
            <param name="infile_pipeline" value="pipeline01"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="C: [1, 10, 100, 1000]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="kernel: ['linear', 'poly', 'rbf', 'sigmoid']"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="k: [3, 5, 7, 9]"/>
                <param name="selected_param_type" value="prep_2_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="with_centering: [True, False]"/>
                <param name="selected_param_type" value="prep_1_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result" >
                <assert_contents>
                    <has_n_columns n="15" />
                    <has_text text="param_preprocessing_1__with_centering"/>
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="selected_search_scheme" value="RandomizedSearchCV"/>
            <param name="infile_pipeline" value="pipeline03"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="n_estimators: np_arange(50, 1001, 50)"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="max_depth: scipy_stats_randint(1, 51)"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="gamma: np_random_uniform(low=0., high=1., size=2)"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="random_state: [324089]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result" >
                <assert_contents>
                    <has_n_columns n="15" />
                    <has_text text="param_estimator__max_depth"/>
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="selected_search_scheme" value="GridSearchCV"/>
            <param name="infile_pipeline" value="pipeline04"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="random_state: list(range(100, 1001, 100))"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="estimator: [ensemble_ExtraTreesClassifier(n_estimators=100, random_state=324089)]"/>
                <param name="selected_param_type" value="prep_1_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result">
                <assert_contents>
                    <has_n_columns n="13"/>
                    <has_text text="0.05363984674329502"/>
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="selected_search_scheme" value="GridSearchCV"/>
            <param name="infile_pipeline" value="pipeline01"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="C: [1, 10, 100, 1000]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_estimator" file="searchCV01" compare="sim_size" delta="1"/>
        </test>
        <test>
            <param name="selected_search_scheme" value="GridSearchCV"/>
            <param name="infile_pipeline" value="pipeline06"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="n_estimators: [10, 50, 200, 1000]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="random_state: [324089]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result">
                <assert_contents>
                    <has_n_columns n="13"/>
                    <has_text_matching expression=".+0.7772355090078996[^/w]+1000[^/d]" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="selected_search_scheme" value="GridSearchCV"/>
            <param name="infile_pipeline" value="pipeline07"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="n_estimators: [10, 50, 100, 200]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="random_state: [324089]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="gamma: [1.0, 2.0]"/>
                <param name="selected_param_type" value="prep_1_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result">
                <assert_contents>
                    <has_n_columns n="14"/>
                    <has_text_matching expression=".+0.05747126436781609[^/d]" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="selected_search_scheme" value="GridSearchCV"/>
            <param name="infile_pipeline" value="pipeline08"/>
            <conditional name="search_param_selector">
                <param name="search_p" value="n_estimators: [10, 50, 100, 200]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="random_state: [324089]"/>
                <param name="selected_param_type" value="final_estimator_p"/>
            </conditional>
            <conditional name="search_param_selector">
                <param name="search_p" value="linkage: ['ward', 'complete', 'average']"/>
                <param name="selected_param_type" value="prep_1_p"/>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular"/>
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns"/>
            <param name="infile2" value="regression_y.tabular" ftype="tabular"/>
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns"/>
            <output name="outfile_result">
                <assert_contents>
                    <has_text_matching expression=".+0.08045977011494253[^/w]+10[^/w]" />
                </assert_contents>
            </output>
        </test>
    </tests>
    <help>
        <![CDATA[
**What it does**
Searches optimized parameter values for an estimator or pipeline through either exhaustive grid cross validation search or Randomized cross validation search.
please refer to `Scikit-learn model_selection GridSearchCV`_, `Scikit-learn model_selection RandomizedSearchCV`_ and `Tuning hyper-parameters`_.

**How to choose search patameters?**

Please refer to `svm`_, `linear_model`_, `ensemble`_, `naive_bayes`_, `tree`_, `neighbors`_ and `xgboost`_ for estimator parameters.
Refer to `sklearn.preprocessing`_, `feature_selection`_, `decomposition`_, `kernel_approximation`_ and `cluster.FeatureAgglomeration`_ for parameter in the pre-processing steps.

**Search parameter input** accepts parameter and setting in key:value pair. One pair per input box. Setting can be list, numpy array, or distribution.
The evaluation of settings supports operations in Math, list comprehension, numpy.arange(np_arange), most numpy.random(e.g., np_random_uniform) and some scipy.stats(e.g., scipy_stats_zipf) classes or functions, and others.

**Examples:**

- K: [3, 5, 7, 9]

- n_estimators: list(range(50, 1001, 50))

- gamma: np_arange(0.01, 1, 0.1)

- alpha: np_random_choice(list(range(1, 51)) + [None], size=20)

- max_depth: scipy_stats_randin(1, 11)

- estimator: [ensemble_ExtraTreesClassifier(n_estimators=100, random_state=324089)]


.. _`Scikit-learn model_selection GridSearchCV`: http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html
.. _`Scikit-learn model_selection RandomizedSearchCV`: http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.RandomizedSearchCV.html
.. _`Tuning hyper-parameters`: http://scikit-learn.org/stable/modules/grid_search.html

.. _`svm`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.svm
.. _`linear_model`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.linear_model
.. _`ensemble`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.ensemble
.. _`naive_bayes`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.naive_bayes
.. _`tree`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.tree
.. _`neighbors`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.neighbors
.. _`xgboost`: https://xgboost.readthedocs.io/en/latest/python/python_api.html

.. _`sklearn.preprocessing`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.preprocessing
.. _`feature_selection`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.feature_selection
.. _`decomposition`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.decomposition
.. _`kernel_approximation`: http://scikit-learn.org/stable/modules/classes.html#module-sklearn.kernel_approximation
.. _`cluster.FeatureAgglomeration`: http://scikit-learn.org/stable/modules/generated/sklearn.cluster.FeatureAgglomeration.html

        ]]>
    </help>
    <expand macro="sklearn_citation"/>
</tool>