Python StratifiedSamplingBinSelector示例

示例#1

文件： test_bin_selection.py 项目： recurve-methods/comparison_groups

def test_plot_records_based_equiv_average_chisquare(
        df_treatment, df_pool, col_name, equivalence_feature_ids,
        equivalence_feature_matrix):
    df_treatment["col2"] = df_treatment[col_name] * 2
    df_treatment["col3"] = df_treatment[col_name] * 3

    df_pool["col2"] = df_pool[col_name] * 2
    df_pool["col3"] = df_pool[col_name] * 3

    stratified_sampling_obj = StratifiedSampling()
    stratified_sampling_obj.add_column(col_name)
    stratified_sampling_obj.add_column("col2")
    stratified_sampling_obj.add_column("col3")

    bin_selection = StratifiedSamplingBinSelector(
        stratified_sampling_obj,
        df_treatment,
        df_pool,
        min_n_bins=2,
        max_n_bins=3,
        random_seed=1,
        equivalence_method='chisquare',
        equivalence_feature_ids=equivalence_feature_ids,
        equivalence_feature_matrix=equivalence_feature_matrix)
    bin_selection.plot_records_based_equiv_average(plot=False)
    results = bin_selection.results_as_json()
    assert 'bins_selected_str' in list(results['n_bin_results'][0].keys())

示例#2

文件： test_bin_selection.py 项目： recurve-methods/comparison_groups

def test_stratified_sampling_fit_and_sample_records_equivalence_idempotent_check(
        df_treatment, df_pool, col_name, equivalence_feature_ids,
        equivalence_feature_matrix):
    df_treatment["col2"] = df_treatment[col_name] * 2
    df_treatment["col3"] = df_treatment[col_name] * 3

    df_pool["col2"] = df_pool[col_name] * 2
    df_pool["col3"] = df_pool[col_name] * 3

    stratified_sampling_obj = StratifiedSampling()
    stratified_sampling_obj.add_column(col_name)
    stratified_sampling_obj.add_column("col2")
    stratified_sampling_obj.add_column("col3")

    StratifiedSamplingBinSelector(
        stratified_sampling_obj,
        df_treatment,
        df_pool,
        min_n_bins=2,
        max_n_bins=3,
        random_seed=1,
        equivalence_method='chisquare',
        equivalence_feature_ids=equivalence_feature_ids,
        equivalence_feature_matrix=equivalence_feature_matrix)
    sample1 = stratified_sampling_obj.data_sample.df.index.values

    stratified_sampling_obj = StratifiedSampling()
    stratified_sampling_obj.add_column(col_name)
    stratified_sampling_obj.add_column("col2")
    stratified_sampling_obj.add_column("col3")
    StratifiedSamplingBinSelector(
        stratified_sampling_obj,
        df_treatment,
        df_pool,
        min_n_bins=2,
        max_n_bins=3,
        random_seed=1,
        equivalence_method='chisquare',
        equivalence_feature_ids=equivalence_feature_ids,
        equivalence_feature_matrix=equivalence_feature_matrix)
    sample2 = stratified_sampling_obj.data_sample.df.index.values
    assert set(sample1) == set(sample2)

示例#3

文件： test_bin_selection.py 项目： recurve-methods/comparison_groups

def test_plot_records_based_equiv_average(df_treatment, df_pool, col_name,
                                          equivalence_feature_ids,
                                          equivalence_feature_matrix):
    df_treatment["col2"] = df_treatment[col_name] * 2
    df_treatment["col3"] = df_treatment[col_name] * 3

    df_pool["col2"] = df_pool[col_name] * 2
    df_pool["col3"] = df_pool[col_name] * 3

    stratified_sampling_obj = StratifiedSampling()
    stratified_sampling_obj.add_column(col_name)
    stratified_sampling_obj.add_column("col2")
    stratified_sampling_obj.add_column("col3")

    bin_selection = StratifiedSamplingBinSelector(
        stratified_sampling_obj,
        df_treatment,
        df_pool,
        min_n_bins=2,
        max_n_bins=3,
        random_seed=1,
        equivalence_method='euclidean',
        equivalence_feature_ids=equivalence_feature_ids,
        equivalence_feature_matrix=equivalence_feature_matrix)
    bin_selection.plot_records_based_equiv_average(plot=False)
    bin_selection.results_as_json()

示例#4

文件： test_bin_selection.py 项目： recurve-methods/comparison_groups

def test_stratified_sampling_fit_and_sample_records_equivalence_too_many_bins(
        df_treatment, df_pool, col_name, equivalence_feature_ids,
        equivalence_feature_matrix):
    stratified_sampling_obj = StratifiedSampling()

    stratified_sampling_obj.add_column(col_name)
    ## attempting to estimate both n_bins and n_samples
    with pytest.raises(ModelSamplingException):
        model_w_selected_bins = StratifiedSamplingBinSelector(
            stratified_sampling_obj,
            df_treatment,
            df_pool,
            min_n_bins=1000,
            max_n_bins=1002,
            random_seed=1,
            equivalence_method='chisquare',
            relax_n_samples_approx_constraint=False,
            equivalence_feature_ids=equivalence_feature_ids,
            equivalence_feature_matrix=equivalence_feature_matrix)

示例#5

文件： test_bin_selection.py 项目： recurve-methods/comparison_groups

def test_stratified_sampling_fit_and_sample_records_equivalence(
        df_treatment, df_pool, col_name, equivalence_feature_ids,
        equivalence_feature_matrix):
    stratified_sampling_obj = StratifiedSampling()
    df_pool["col2"] = df_pool[col_name]
    df_treatment["col2"] = df_treatment[col_name]
    stratified_sampling_obj.add_column(col_name)
    stratified_sampling_obj.add_column("col2")
    ## attempting to estimate both n_bins and n_samples
    StratifiedSamplingBinSelector(
        stratified_sampling_obj,
        df_treatment,
        df_pool,
        min_n_bins=4,
        max_n_bins=6,
        random_seed=1,
        equivalence_method='chisquare',
        equivalence_feature_ids=equivalence_feature_ids,
        equivalence_feature_matrix=equivalence_feature_matrix)
    output = stratified_sampling_obj.data_sample.df
    bins_df = stratified_sampling_obj.diagnostics().count_bins()