def create_results_object_model()

in experimenter/experimenter/jetstream/models.py [0:0]

• 63 lines of code
• 14 McCabe index (conditional complexity)

def create_results_object_model(data: JetstreamData):
    branches = {data_point.branch for data_point in data}

    # create a dynamic model with all branches, leveraging BranchComparisonData
    PairwiseBranchComparisonData = create_model(
        "PairwiseBranchComparisonData",
        **{
            branch: (BranchComparisonData, Field(default_factory=BranchComparisonData))
            for branch in branches
        },
    )

    # create a dynamic model with all branches, leveraging SignificanceData
    PairwiseSignificanceData = create_model(
        "PairwiseSignificanceData",
        **{
            branch: (SignificanceData, Field(default_factory=SignificanceData))
            for branch in branches
        },
    )

    class PairwiseMetricData(MetricData):
        difference: PairwiseBranchComparisonData = Field(
            default_factory=PairwiseBranchComparisonData
        )
        relative_uplift: PairwiseBranchComparisonData = Field(
            default_factory=PairwiseBranchComparisonData
        )
        significance: PairwiseSignificanceData = Field(
            default_factory=PairwiseSignificanceData
        )

    metrics = {data_point.metric for data_point in data}

    # Dynamically create our grouped models which are dependent on metrics
    # available for a given experiment
    SearchData = create_model(
        "SearchData",
        **{
            metric_name: (PairwiseMetricData, Field(default_factory=PairwiseMetricData))
            for metric_name in metrics
            if metric_name in SEARCH_METRICS
        },
    )

    UsageData = create_model(
        "UsageData",
        **{
            metric_name: (PairwiseMetricData, Field(default_factory=PairwiseMetricData))
            for metric_name in metrics
            if metric_name in USAGE_METRICS
        },
    )

    OtherData = create_model(
        "OtherData",
        **{
            metric_name: (PairwiseMetricData, Field(default_factory=PairwiseMetricData))
            for metric_name in metrics
            if metric_name not in SEARCH_METRICS + USAGE_METRICS
        },
    )

    class BranchData(BaseModel):
        search_metrics: SearchData = Field(default_factory=SearchData)
        usage_metrics: UsageData = Field(default_factory=UsageData)
        other_metrics: OtherData = Field(default_factory=OtherData)

    class Branch(BaseModel):
        is_control: bool = False
        branch_data: BranchData = Field(default_factory=BranchData)

    # Create ResultsObjectModel model which is dependent on
    # branches available for a given experiment
    return create_model(
        "ResultsObjectModel",
        **{branch: (Branch, Field(default_factory=Branch)) for branch in branches},
        __base__=ResultsObjectModelBase,
    )