Source code for message_ix_models.model.macro

"""Tools for calibrating MACRO for MESSAGEix-GLOBIOM.

See :doc:`message-ix:macro` for *general* documentation on MACRO and MESSAGE-MACRO. This
module contains tools specifically for using these models with MESSAGEix-GLOBIOM.
"""

import logging
from functools import lru_cache
from itertools import product
from pathlib import Path
from typing import TYPE_CHECKING, List, Literal, Mapping, Optional, Union

import pandas as pd

from message_ix_models.model.bare import get_spec
from message_ix_models.util import nodes_ex_world

if TYPE_CHECKING:
    from sdmx.model.v21 import Code

    from message_ix_models import Context

log = logging.getLogger(__name__)

#: Default set of commodities to include in :func:`generate`.
COMMODITY = ["i_therm", "i_spec", "rc_spec", "rc_therm", "transport"]


[docs]def generate( parameter: Literal["aeei", "config", "depr", "drate", "lotol"], context: "Context", commodities: Union[List[str], List["Code"]] = COMMODITY, value: Optional[float] = None, ) -> pd.DataFrame: """Generate uniform data for one :mod:`message_ix.macro` `parameter`. :meth:`message_ix.Scenario.add_macro` expects as its `data` parameter a :class:`dict` that maps certain MACRO parameter names (or the special name "config") to :class:`.pandas.DataFrame`. This function generates data for those data frames. For the particular dimensions, generate automatically includes: - "node": All nodes in the node code list given by :func:`.nodes_ex_world`, for the node list indicated by :attr:`.model.Config.regions`. - "year": All periods from the period *before* the first model year. - "commodity": The elements of `commodities`. - "sector": If each entry of `commodities` is a :class:`.Code` and has an annotation with id="macro-sector", the value of that annotation. Otherwise, the same as `commodity`. `value` supplies the parameter value, which is the same for all observations. The labels level="useful" and unit="-" are fixed. Parameters ---------- parameter : str MACRO parameter for which to generate data. context Used with :func:`.bare.get_spec`. commodities : list of str or .Code Commodities to include in the MESSAGE-MACRO linkage. value : float Parameter value. Returns ------- pandas.DataFrame The columns vary according to `parameter`: - "aeei": node, sector, year, value, unit. - "depr", "drate", or "lotol": node, value, unit. - "config": node, sector, commodity, level, year. """ spec = get_spec(context) if isinstance(commodities[0], str): c_codes = spec.add.set["commodity"] else: c_codes = commodities @lru_cache def _sector(commodity: str) -> str: try: idx = c_codes.index(commodity) return str(c_codes[idx].get_annotation(id="macro-sector").text) except (KeyError, ValueError) as e: log.info(e) return str(commodity) # AEEI data must begin from the period before the first model period y0_index = spec.add.set["year"].index(spec.add.y0) iterables = dict( c_s=zip( # Paired commodity and sector map(str, commodities), map(_sector, commodities) ), level=["useful"], node=nodes_ex_world(spec.add.N), sector=map(_sector, commodities), year=spec.add.set["year"][y0_index:], ) if parameter == "aeei": dims = ["node", "year", "sector"] iterables.update(year=spec.add.set["year"][y0_index - 1 :]) elif parameter == "config": dims = ["node", "c_s", "level", "year"] assert value is None elif parameter in ("depr", "drate", "lotol"): dims = ["node"] else: raise NotImplementedError(f"generate(…) for MACRO parameter {parameter!r}") result = pd.DataFrame( [tuple(values) for values in product(*[iterables[d] for d in dims])], columns=dims, ) if parameter == "config": return pd.concat( [ result.drop("c_s", axis=1), pd.DataFrame(result["c_s"].tolist(), columns=["commodity", "sector"]), ], axis=1, ) else: return result.assign(value=value, unit="-")
[docs]def load(base_path: Path) -> Mapping[str, pd.DataFrame]: """Load MACRO data from CSV files. The function reads files in the simple/long CSV format understood by :func:`.genno.operator.load_file`. For use with :meth:`~message_ix.Scenario.add_macro`, the dimension names should be given in full, for instance "node" or "sector". Parameters ---------- base_path : pathlib.Path Directory containing zero or more CSV files. Returns ------- dict of (str -> pandas.DataFrame) Mapping from MACRO calibration parameter names to data; one entry for each file in `base_path`. """ from genno.operator import load_file result = {} for filename in base_path.glob("*.csv"): name = filename.stem q = load_file(filename, name=name) result[name] = ( q.to_frame() .reset_index() .rename(columns={name: "value"}) .assign(unit=f"{q.units:~}" or "-") ) return result