Merge pull request #72 from brightway-lca/refactored_dynamic_characte…

…rization Refactored dynamic characterization
brightway-lca · Jul 17, 2024 · f01c5b0 · f01c5b0
2 parents 78c68fe + 89af837
commit f01c5b0
Show file tree

Hide file tree

Showing 11 changed files with 217 additions and 973 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,3 +1,6 @@
+## v0.1.8 (2024-07-17)
+* Moved dynamic characterization functionality completely to [dynamic_characterization](https://github.com/brightway-lca/dynamic_characterization). In the course of this, it was dynamic characterization was updated and is much faster now. See also https://github.com/brightway-lca/dynamic_characterization/pull/3
+
 ## v0.1.7 (2024-07-11)
 * Fixed some dependencies
 
@@ -6,7 +9,7 @@
 * Added option to calculate the dynamic LCI directly from the timeline without expanding the technosphere matrix
 
 ## v0.1.5 (2024-06-28)
-* Refactored dynamic characterization to separate package [dynamic_characterization](https://github.com/TimoDiepers/dynamic_characterization)
+* Refactored dynamic characterization to separate package [dynamic_characterization](https://github.com/brightway-lca/dynamic_characterization)
 
 ## v0.1.4 (2024-06-15)
 * Handled emissions occuring outside of fixed time horizon in dynamic characterization [#46](https://github.com/brightway-lca/bw_timex/issues/46)

diff --git a/bw_timex/__init__.py b/bw_timex/__init__.py
@@ -1,9 +1,8 @@
 from .dynamic_biosphere_builder import DynamicBiosphereBuilder
-from .dynamic_characterization import DynamicCharacterization
 from .edge_extractor import EdgeExtractor
 from .helper_classes import SetList
 from .matrix_modifier import MatrixModifier
 from .timeline_builder import TimelineBuilder
 from .timex_lca import TimexLCA
 
-__version__ = "0.1.7"
+__version__ = "0.1.8"
diff --git a/bw_timex/dynamic_characterization.py b/bw_timex/dynamic_characterization.py
diff --git a/bw_timex/timex_lca.py b/bw_timex/timex_lca.py
@@ -23,12 +23,12 @@
 from bw2data.backends.schema import ActivityDataset as AD
 from bw2data.backends.schema import get_id
 from bw2data.errors import Brightway2Project
+from dynamic_characterization import characterize_dynamic_inventory
 from peewee import fn
 from scipy import sparse
 
 # from .dynamic_biosphere_builder import DynamicBiosphereBuilder
 from .dynamic_biosphere_builder import DynamicBiosphereBuilder
-from .dynamic_characterization import DynamicCharacterization
 from .helper_classes import SetList, TimeMappingDict
 from .matrix_modifier import MatrixModifier
 from .timeline_builder import TimelineBuilder
@@ -193,9 +193,13 @@ def build_timeline(
         self.cutoff = cutoff
         self.max_calc = max_calc
 
-        self.add_static_activities_to_time_mapping_dict()  # pre-populate the activity time mapping dict with the static activities. Doing this here because we need the temporal grouping for consistent times resolution.
+        # pre-populate the activity time mapping dict with the static activities.
+        # Doing this here because we need the temporal grouping for consistent times resolution.
+        self.add_static_activities_to_time_mapping_dict()
 
-        # Create timeline builder that does the graph traversal (similar to bw_temporalis) and extracts all edges with their temporal information. Can later be used to build a timeline with the TimelineBuilder.build_timeline() method.
+        # Create timeline builder that does the graph traversal (similar to bw_temporalis) and extracts
+        # all edges with their temporal information. Can later be used to build a timeline with the
+        # TimelineBuilder.build_timeline() method.
         self.timeline_builder = TimelineBuilder(
             self.static_lca,
             self.edge_filter_function,
@@ -322,11 +326,12 @@ def static_lcia(self) -> None:
 
     def dynamic_lcia(
         self,
-        metric: str | None = "GWP",
-        time_horizon: int | None = 100,
-        fixed_time_horizon: bool | None = False,
+        metric: str = "radiative_forcing",
+        time_horizon: int = 100,
+        fixed_time_horizon: bool = False,
+        time_horizon_start: datetime = None,
         characterization_function_dict: dict = None,
-        cumsum: bool | None = True,
+        characterization_function_co2: dict = None,
     ) -> pd.DataFrame:
         """
         Calculates dynamic LCIA with the `DynamicCharacterization` class using the dynamic inventory and dynamic
@@ -347,15 +352,18 @@ def dynamic_lcia(
         Parameters
         ----------
         metric : str, optional
-            the metric for which the dynamic LCIA should be calculated. Default is "GWP". Available: "GWP" and "radiative_forcing"
+            the metric for which the dynamic LCIA should be calculated. Default is "radiative_forcing". Available: "GWP" and "radiative_forcing"
         time_horizon: int, optional
             the time horizon for the impact assessment. Unit is years. Default is 100.
         fixed_time_horizon: bool, optional
             Whether the emission time horizon for all emissions is calculated from the functional unit (fixed_time_horizon=True) or from the time of the emission (fixed_time_horizon=False). Default is False.
+        time_horizon_start: pd.Timestamp, optional
+            The starting timestamp of the time horizon for the dynamic characterization. Only needed for fixed time horizons. Default is datetime.now().
         characterization_function_dict: dict, optional
             Dict of the form {biosphere_flow_database_id: characterization_function}. Default is None, which triggers the use of the provided dynamic characterization functions based on IPCC AR6 Chapter 7.
-        cumsum: bool, optional
-            Whether to calculate the cumulative sum of the characterization results. Default is True.
+        characterization_function_co2: Callable, optional
+            Characterization function for CO2 emissions. Necessary if GWP metrix is chosen. Default is None, which triggers the use of the provided dynamic characterization function of co2 based on IPCC AR6 Chapter 7.
+
 
         Returns
         -------
@@ -372,36 +380,36 @@ def dynamic_lcia(
                 "Dynamic lci not yet calculated. Call TimexLCA.calculate_dynamic_lci() first."
             )
 
+        # Set a default for inventory_in_time_horizon using the full dynamic_inventory_df
+        inventory_in_time_horizon = self.dynamic_inventory_df
+
+        # Calculate the latest considered impact date
         t0_date = pd.Timestamp(self.timeline_builder.edge_extractor.t0.date[0])
         latest_considered_impact = t0_date + pd.DateOffset(years=time_horizon)
-        last_emission = self.dynamic_inventory_df.date.max()
-        if fixed_time_horizon and latest_considered_impact < last_emission:
-            warnings.warn(
-                "An emission occurs outside of the specified time horizon and will not be characterized. Please make sure this is intended."
-            )
-            self.dynamic_inventory_df = self.dynamic_inventory_df[
-                self.dynamic_inventory_df.date <= latest_considered_impact
-            ]
-
-        self.metric = metric
-        self.time_horizon = time_horizon
-        self.fixed_time_horizon = fixed_time_horizon
 
-        self.dynamic_characterizer = DynamicCharacterization(
-            self.dynamic_inventory_df,
-            self.demand_timing_dict,
-            self.temporal_grouping,
-            self.method,
-            characterization_function_dict,
-        )
+        # Update inventory_in_time_horizon if a fixed time horizon is used
+        if fixed_time_horizon:
+            last_emission = self.dynamic_inventory_df.date.max()
+            if latest_considered_impact < last_emission:
+                warnings.warn(
+                    "An emission occurs outside of the specified time horizon and will not be characterized. Please make sure this is intended."
+                )
+                inventory_in_time_horizon = self.dynamic_inventory_df[
+                    self.dynamic_inventory_df.date <= latest_considered_impact
+                ]
 
-        self.characterized_inventory = (
-            self.dynamic_characterizer.characterize_dynamic_inventory(
-                metric,
-                time_horizon,
-                fixed_time_horizon,
-                cumsum,
-            )
+        if not time_horizon_start:
+            time_horizon_start = t0_date
+
+        self.characterized_inventory = characterize_dynamic_inventory(
+            dynamic_inventory_df=inventory_in_time_horizon,
+            metric=metric,
+            characterization_function_dict=characterization_function_dict,
+            base_lcia_method=self.method,
+            time_horizon=time_horizon,
+            fixed_time_horizon=fixed_time_horizon,
+            time_horizon_start=time_horizon_start,
+            characterization_function_co2=characterization_function_co2,
         )
 
         self.dynamic_score = self.characterized_inventory["amount"].sum()
@@ -1153,7 +1161,7 @@ def plot_dynamic_characterized_inventory(
 
         if not hasattr(self, "characterized_inventory"):
             warnings.warn(
-                "Characterized inventory not yet calculated. Call TimexLCA.characterize_dynamic_lci() first."
+                "Characterized inventory not yet calculated. Call TimexLCA.dynamic_lcia() first."
             )
             return
 
@@ -1175,48 +1183,34 @@ def plot_dynamic_characterized_inventory(
             plot_data["amount_sum"] = plot_data["amount"].cumsum()
             plot_data["activity_label"] = "All activities"
 
-        else:
-            plot_data["activity_label"] = plot_data.apply(
-                lambda row: bd.get_activity(
-                    self.activity_time_mapping_dict_reversed[row.activity][0]
-                )["name"],
-                axis=1,
-            )
-        # Plotting
-        plt.figure(figsize=(14, 6))
-        axes = sb.scatterplot(x="date", y=amount, hue="activity_label", data=plot_data)
+        else:  # plotting activities separate
 
-        # Determine the plotting labels and titles based on the characterization method
-        if self.metric == "radiative_forcing":
-            label_legend = "Radiative forcing [W/m2]"
-            title = "Radiative forcing"
-        elif self.metric == "GWP":
-            label_legend = "GWP [kg CO2-eq]"
-            title = "GWP"
+            activity_name_cache = {}
 
-        if self.fixed_time_horizon:
-            suptitle = f" \nTH of {self.time_horizon} years starting at FU,"
-        else:
-            suptitle = f" \nTH of {self.time_horizon} years starting at each emission,"
+            for activity in plot_data["activity"].unique():
+                if activity not in activity_name_cache:
+                    activity_name_cache[activity] = bd.get_activity(
+                        self.activity_time_mapping_dict_reversed[activity][0]
+                    )["name"]
+
+            plot_data["activity_label"] = plot_data["activity"].map(activity_name_cache)
 
-        suptitle += f" temporal resolution of inventories: {self.temporal_grouping}"
+        # Plotting
+        plt.figure(figsize=(14, 6))
+        axes = sb.scatterplot(x="date", y=amount, hue="activity_label", data=plot_data)
 
         # Determine y-axis limit flexibly
         if plot_data[amount].min() < 0:
             ymin = plot_data[amount].min() * 1.1
         else:
             ymin = 0
 
-        axes.set_title(title)
         axes.set_axisbelow(True)
         axes.set_ylim(bottom=ymin)
-        axes.set_ylabel(label_legend)
         axes.set_xlabel("Time")
 
         handles, labels = axes.get_legend_handles_labels()
         axes.legend(handles[::-1], labels[::-1])
-        plt.title(title, fontsize=16, y=1.05)
-        plt.suptitle(suptitle, fontsize=12, y=0.95)
         plt.grid()
         plt.show()
 

diff --git a/bw_timex/utils.py b/bw_timex/utils.py
@@ -1,10 +1,10 @@
 import warnings
+from datetime import datetime
+from typing import Callable, List, Optional, Union
+
+import bw2data as bd
 import matplotlib.pyplot as plt
 import pandas as pd
-import bw2data as bd
-
-from datetime import datetime
-from typing import Callable, Union, List, Optional
 
 
 def extract_date_as_integer(dt_obj: datetime, time_res: Optional[str] = "year") -> int:
@@ -58,7 +58,7 @@ def extract_date_as_string(temporal_grouping: str, timestamp: datetime) -> str:
         Temporal grouping for the date string. Options are: 'year', 'month', 'day', 'hour'
     timestamp : datetime
         Datetime object to be converted to a string.
-    
+
     Returns
     -------
     date_as_string
@@ -92,7 +92,7 @@ def convert_date_string_to_datetime(temporal_grouping, datestring) -> datetime:
         Temporal grouping for the date string. Options are: 'year', 'month', 'day', 'hour'
     datestring : str
         Date as a string
-    
+
     Returns
     -------
     datetime
@@ -126,12 +126,12 @@ def add_flows_to_characterization_function_dict(
     Parameters
     ----------
     flows : Union[str, List[str]]
-        Flow or list of flows to be added to the characterization function dictionary. 
+        Flow or list of flows to be added to the characterization function dictionary.
     func : Callable
         Dynamic characterization function for flow.
     characterization_function_dict : dict, optional
         Dictionary of flows and their corresponding characterization functions. Default is an empty dictionary.
-    
+
     Returns
     -------
     dict
@@ -151,8 +151,7 @@ def add_flows_to_characterization_function_dict(
 
 
 def plot_characterized_inventory_as_waterfall(
-    characterized_inventory,
-    metric,
+    lca_obj,
     static_scores=None,
     prospective_scores=None,
     order_stacked_activities=None,
@@ -163,10 +162,8 @@ def plot_characterized_inventory_as_waterfall(
 
     Parameters
     ----------
-    characterized_inventory : pd.DataFrame
-        Dataframe of dynamic characterized inventory, such as TimexLCA.characterized_inventory
-    metric : str
-        Impact assessment method. Only 'GWP' is supported at the moment.
+    lca_obj : TimexLCA
+        LCA object with characterized inventory data.
     static_scores : dict, optional
         Dictionary of static scores. Default is None.
     prospective_scores : dict, optional
@@ -179,32 +176,29 @@ def plot_characterized_inventory_as_waterfall(
     None but plots the waterfall chart.
 
     """
-    # Check for necessary columns
-    if not {"date", "activity_name", "amount"}.issubset(
-        characterized_inventory.columns
-    ):
-        raise ValueError(
-            "DataFrame must contain 'date', 'activity_name', and 'amount' columns."
-        )
-    if metric != "GWP":
-        raise NotImplementedError(
-            f"Only GWP metric is supported at the moment, not {metric}."
-        )
+    if not hasattr(lca_obj, "characterized_inventory"):
+        raise ValueError("LCA object does not have characterized inventory data.")
+
+    if not hasattr(lca_obj, "activity_time_mapping_dict_reversed"):
+        raise ValueError("Make sure to pass an instance of a TimexLCA.")
 
     # Grouping and summing data
-    plot_data = characterized_inventory.groupby(
-        ["date", "activity_name"], as_index=False
+    plot_data = lca_obj.characterized_inventory.groupby(
+        ["date", "activity"], as_index=False
     ).sum()
     plot_data["year"] = plot_data[
         "date"
     ].dt.year  # TODO make temporal resolution flexible
 
     # Optimized activity label fetching
-    unique_activities = plot_data["activity_name"].unique()
+    unique_activities = plot_data["activity"].unique()
     activity_labels = {
-        activity: bd.get_activity(activity)["name"] for activity in unique_activities
+        idx: bd.get_activity(lca_obj.activity_time_mapping_dict_reversed[idx][0])[
+            "name"
+        ]
+        for idx in unique_activities
     }
-    plot_data["activity_label"] = plot_data["activity_name"].map(activity_labels)
+    plot_data["activity_label"] = plot_data["activity"].map(activity_labels)
     # Pivoting data for plotting
     pivoted_data = plot_data.pivot(
         index="year", columns="activity_label", values="amount"

diff --git a/notebooks/example_setac.ipynb b/notebooks/example_setac.ipynb
@@ -1683,8 +1683,7 @@
     ")\n",
     "\n",
     "plot_characterized_inventory_as_waterfall(\n",
-    "    tlca.characterized_inventory,\n",
-    "    metric=tlca.metric,\n",
+    "    tlca,\n",
     "    static_scores=static_scores,\n",
     "    prospective_scores=prospective_scores,\n",
     "    order_stacked_activities=order_stacked_activities,\n",