analysis.py

import numpy as np
import pandas as pd

import preprocess as pp
import beam_search as bs
import distribution_false_discoveries as dfd

def analysis(data_name=None, trend_name=None, model_params=None, beam_search_params=None, constraints=None, dfd_params=None, wcs_params=None):

    dataset, attributes, descriptives = pp.preprocess(data_name=data_name, trend_name=trend_name)
    print(descriptives)
    print(attributes)
    print(dataset.dtypes)

    beam_search_params['pareto'] = False

    # check if distribution has to be made
    if dfd_params['make']:
        # build dfd, as a pd.DataFrame where the quality values are a list, and other values are distribution params
        distribution_params = dfd.distribution_false_discoveries_params(m=dfd_params['m'], model_params=model_params,
                                                                        beam_search_params=beam_search_params,
                                                                        dataset=dataset, attributes=attributes, descriptives=descriptives, 
                                                                        wcs_params=wcs_params, constraints=constraints)

    else:
        distribution_params = None               

    # a single run
    print(beam_search_params)
    result_emm, general_params, considered_subgroups = bs.beam_search(dataset=dataset, attributes=attributes, descriptives=descriptives, 
                                                                      model_params=model_params, beam_search_params=beam_search_params, 
                                                                      wcs_params=wcs_params, constraints=constraints)

    result_analysis = result_emm

    return result_analysis, general_params, considered_subgroups, distribution_params