class Tabulation:
    """Model post-processing class for generating two outputs:
    1. Weekly wholesaler, product-pack and vehicle level spends, incremental revenue along with the ROIs
    2. Dataframes for visualizing diminishing return curves for each wholesaler, product-pack and vehicle
    """

    def __init__(self, dataset_factory: "DataFactoryType", encodings: dict, gt_model: "TrainableModule"):
        """Class initialization needed for model post-processing

        Args:
            dataset_factory (DataFactoryType): generator function for iterating over
                                                                          batches of data
            encodings (dict): mapping from names to indices for wholesalers, brands, products, vehicles and so on
            gt_model (EconomicNetwork): trained ground truth model to be used for inference
        """
        self.dataset_factory = dataset_factory
        self.encodings = encodings
        self.gt_model = gt_model

    def get_weekly_roi_facts(self):
        """Generates weekly wholesaler, product-pack and vehicle level spends, incremental revenues and ROIs

        Returns:
            pd.DataFrame: weekly roi facts for each wholesaler, product-pack and vehicle
        """
        # apply_sales_mask is set to True because the instantaneous impacts learnt by the model are masked out
        # in betagamma layer, so the impacts crucial for calculating ROIs are masked out.
        # So, investments need to masked too fro the ROI numbers to make sense.
        all_investments_df = get_all_investments_df(
            self.dataset_factory,
            self.encodings,
            denormalize=True,
            apply_sales_mask=True,
            index_types=GRANULARITY_OPTIONS,
        )
        self.all_investments = stack_df_and_rename_col(all_investments_df, "spend")
        all_inc_revs_df = get_all_inc_revenues_df(
            self.dataset_factory, self.encodings, self.gt_model, denormalize=True, total_impact_from_date=True
        )
        all_inc_revs_df.index = all_investments_df.index
        self.all_inc_revs = stack_df_and_rename_col(all_inc_revs_df, "inc_revenue")
        weekly_roi_facts = self.all_investments.merge(
            self.all_inc_revs,
            on=(self.all_investments.columns[:-1].tolist()),
        )
        weekly_roi_facts["roi"] = weekly_roi_facts["inc_revenue"] / (weekly_roi_facts["spend"] + EPSILON)
        return weekly_roi_facts

    def get_roicurves(self):
        """Generates dataframes for visualizing diminishing return curves for each wholesaler, product-pack and vehicle

        Returns:
            pd.DataFrame, pd.DataFrame: diminishing return curves for media vehicles
        """
        vehicles = get_lookups(self.encodings["vehicle"])
        vehicle_roicurves_df = get_all_roicurves_df(
            self.dataset_factory, self.encodings, self.gt_model, vehicles, denormalize=True
        )
        return vehicle_roicurves_df