add boundary filtering to config

config/base.toml

@@ -4,6 +4,10 @@ region = "leeds"
 number_of_households = 10000
 zone_id = "OA21CD"
 travel_times = true
+boundary_geography = "OA"
+boundary_filter_column = "LEP22NM1"
+boundary_filter_values = ["Leeds City Region"]
+
 
 [work_assignment]
 use_percentages = true

config/base_500.toml

@@ -4,6 +4,10 @@ region = "leeds"
 number_of_households = 500
 zone_id = "OA21CD"
 travel_times = true
+boundary_geography = "OA"
+boundary_filter_column = "LEP22NM1"
+boundary_filter_values = ["Leeds City Region"]
+
 
 [work_assignment]
 use_percentages = true

config/base_5000.toml

@@ -4,6 +4,10 @@ region = "leeds"
 number_of_households = 5000
 zone_id = "OA21CD"
 travel_times = true
+boundary_geography = "OA"
+boundary_filter_column = "LEP22NM1"
+boundary_filter_values = ["Leeds City Region"]
+
 
 [work_assignment]
 use_percentages = true

config/base_all.toml

@@ -3,6 +3,10 @@ seed = 0
 region = "leeds"
 zone_id = "OA21CD"
 travel_times = true
+boundary_geography = "OA"
+boundary_filter_column = "LEP22NM1"
+boundary_filter_values = ["Leeds City Region"]
+
 
 [work_assignment]
 use_percentages = false

scripts/0_preprocess_inputs.py

@@ -1,47 +1,60 @@
 import geopandas as gpd
 
 import acbm
-from acbm.logger_config import preprocessing_logger_logger as logger
+from acbm.cli import acbm_cli
+from acbm.config import load_config
+from acbm.logger_config import preprocessing_logger as logger
 from acbm.preprocessing import edit_boundary_resolution, filter_boundaries
 
-# ----- BOUNDARIES
-logger.info("Preprocessing Boundary Layer")
 
-## Read in the boundary layer for the whole of England
+@acbm_cli
+def main(config_file):
+ config = load_config(config_file)
+ config.init_rng()
 
-logger.info("1. Reading in the boundary layer for the whole of England")
+ # ----- BOUNDARIES
+ logger.info("Preprocessing Boundary Layer")
 
+ ## Read in the boundary layer for the whole of England
 
-boundaries = gpd.read_file(
- acbm.root_path / "data/external/boundaries/oa_england.geojson"
-)
+ logger.info("1. Reading in the boundary layer for the whole of England")
 
-boundaries = boundaries.to_crs(epsg=4326)
+ boundaries = gpd.read_file(
+ acbm.root_path / "data/external/boundaries/oa_england.geojson"
+ )
 
-## Dissolve boundaries if resolution is MSOA
+ boundaries = boundaries.to_crs(epsg=4326)
 
-boundary_geography = "OA" # can only be OA or MSOA
-logger.info(f"2. Dissolving boundaries to {boundary_geography} level")
+ ## Dissolve boundaries if resolution is MSOA
 
-boundaries = edit_boundary_resolution(boundaries, boundary_geography)
+ boundary_geography = config.parameters.boundary_geography # can only be OA or MSOA
+ logger.info(f"2. Dissolving boundaries to {boundary_geography} level")
 
+ boundaries = edit_boundary_resolution(boundaries, boundary_geography)
 
-## Filter to study area
+ ## Filter to study area
 
-logger.info("3. Filtering boundaries to specified study area")
-# TODO get from config and log
-# logger.info(f"3. Filtering boundaries to {config.parameters.boundary_filter_column} = {config.parameters.study_area}")
+ logger.info("3. Filtering boundaries to specified study area")
+ # TODO get from config and log
+ # logger.info(f"3. Filtering boundaries to {config.parameters.boundary_filter_column} = {config.parameters.study_area}")
 
-boundaries_filtered = filter_boundaries(
- boundaries=boundaries, column="LEP22NM1", values=["Leeds City Region"]
-)
+ boundaries_filtered = filter_boundaries(
+ # boundaries=boundaries, column="LEP22NM1", values=["Leeds City Region"]
+ boundaries=boundaries,
+ column=config.parameters.boundary_filter_column,
+ values=config.parameters.boundary_filter_values,
+ )
 
-## Save the output as parquet
-logger.info(
- f"4. Saving the boundaries to {acbm.root_path / 'data/external/boundaries/'} path"
-)
+ ## Save the output as parquet
+ logger.info(
+  f"4. Saving the boundaries to {acbm.root_path / 'data/external/boundaries/'} path"
+ )
 
-boundaries_filtered.to_file(
- acbm.root_path / "data/external/boundaries/study_area_zones.geojson",
- driver="GeoJSON",
-)
+ boundaries_filtered.to_file(
+ acbm.root_path / "data/external/boundaries/study_area_zones.geojson",
+ driver="GeoJSON",
+ )
+
+
+if __name__ == "__main__":
+ main()

scripts/run_pipeline.sh

@@ -2,7 +2,7 @@
 
 set -e
 
-python scripts/0_preprocessing_inputs.py --config_file $1
+python scripts/0_preprocess_inputs.py --config_file $1
 python scripts/1_prep_synthpop.py --config_file $1
 python scripts/2_match_households_and_individuals.py --config_file $1
 python scripts/3.1_assign_primary_feasible_zones.py --config_file $1

src/acbm/config.py

@@ -13,6 +13,9 @@ class Parameters(BaseModel):
  number_of_households: int | None = None
  zone_id: str
  travel_times: bool
+ boundary_geography: str
+ boundary_filter_column: str
+ boundary_filter_values: list[str]
 
 
 @dataclass(frozen=True)