MSFTv4 model runs

Author: collijk

poetry.lock

@@ -44,6 +44,9 @@ scikit-learn = "^1.6.0"
 pyarrow = "^18.1.0"
 torch = "^2.5.1"
 lightning = "^2.5.0.post0"
+scikit-image = "^0.25.1"
+pypdf2 = "^3.0.1"
+contextily = "^1.6.2"
 
 [tool.poetry.group.dev.dependencies]
 mkdocstrings = {version = "*", extras = ["python"]}
@@ -60,6 +63,9 @@ pytest-github-actions-annotate-failures = "*"
 pytest-cov = "*"
 python-kacl = "*"
 ruff = "*"
+pandas-stubs = "^2.2.3.241126"
+types-tqdm = "^4.67.0.20241221"
+types-pyyaml = "^6.0.12.20241230"
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]
@@ -143,17 +149,24 @@ exclude_lines = [
 # This is the global mypy configuration.
 # Avoid changing this!
 strict = true  # See all the enabled flags `mypy --help | grep -A 10 'Strict mode'`
-disallow_any_unimported = true
+disallow_any_unimported = false
 
 # If you need to ignore something for some specific module,
 # add overrides for them. Avoid changing the global config!
 # For example:
-# [[tool.mypy.overrides]]
-# module = [
-#  "my_unpyted_dependency1.*",
-#  "my_unpyted_dependency2.*"
-# ]
-# ignore_missing_imports = true
+[[tool.mypy.overrides]]
+module = [
+    "geopandas.*",
+    "shapely.*",
+    "seaborn.*",
+    "mpl_toolkits.*",
+    "rasterio.*",
+    "affine.*",
+    "scipy.*",
+    "sklearn.*",
+    "contextily.*",
+]
+ignore_missing_imports = true
 
 # [[tool.mypy.overrides]]
 # module = [

pyproject.toml

@@ -42,4 +42,4 @@ def task_runner() -> None:
     for name, runner in task_runners.items():
         task_runner.add_command(runner, name)
 
-    pmtask.add_command(task_runner)
+    pmtask.add_command(task_runner)

src/rra_population_model/cli.py

@@ -18,8 +18,8 @@
     with_queue,
     with_verbose,
 )
-from rra_population_model import constants as pmc
 
+from rra_population_model import constants as pmc
 
 _T = TypeVar("_T")
 _P = ParamSpec("_P")
@@ -67,6 +67,7 @@ def with_year(
         help="Year to run.",
     )
 
+
 def with_time_point(
     choices: Collection[str] = pmc.ALL_TIME_POINTS,
     *,
@@ -91,6 +92,7 @@ def with_block_key() -> ClickOption[_P, _T]:
         help="Block key of block to run.",
     )
 
+
 def with_tile_key() -> ClickOption[_P, _T]:
     return click.option(
         "--tile-key",
@@ -100,6 +102,7 @@ def with_tile_key() -> ClickOption[_P, _T]:
         help="Tile key of tile to run.",
     )
 
+
 def with_model_name() -> ClickOption[_P, _T]:
     return click.option(
         "--model-name",
@@ -115,20 +118,20 @@ def with_model_name() -> ClickOption[_P, _T]:
     "ClickOption",
     "convert_choice",
     "process_choices",
+    "with_block_key",
     "with_choice",
     "with_debugger",
     "with_dry_run",
     "with_input_directory",
+    "with_iso3",
     "with_num_cores",
     "with_output_directory",
     "with_overwrite",
     "with_progress_bar",
     "with_queue",
-    "with_verbose",
     "with_resolution",
-    "with_time_point",
-    "with_block_key",
     "with_tile_key",
+    "with_time_point",
+    "with_verbose",
     "with_year",
-    "with_iso3",
-]
+]

src/rra_population_model/cli_options.py

@@ -1,9 +1,9 @@
 import itertools
+import warnings
 from pathlib import Path
-import pyproj
 
+import pyproj
 from pydantic import BaseModel, model_validator
-import warnings
 
 RRA_ROOT = Path("/mnt/team/rapidresponse/")
 RRA_CREDENTIALS_ROOT = RRA_ROOT / "priv" / "shared" / "credentials"
@@ -16,14 +16,19 @@
 RESOLUTIONS = ["40", "100", "250", "500", "1000"]
 
 FEATURE_AVERAGE_RADII = [
-    100, 500, 1000, 2500, 5000, 10000,
+    100,
+    500,
+    1000,
+    2500,
+    5000,
+    10000,
 ]
 
 GHSL_TIME_POINTS = [f"{y}q1" for y in range(1975, 2024)]
 MICROSOFT_TIME_POINTS = {
-    "microsoft_v2": [f"{y}q{q}" for q, y in itertools.product(range(1, 5), range(2018, 2024))][
-        :-1
-    ],
+    "microsoft_v2": [
+        f"{y}q{q}" for q, y in itertools.product(range(1, 5), range(2018, 2024))
+    ][:-1],
     "microsoft_v3": ["2023q3"],
     "microsoft_v4": ["2023q4"],
 }
@@ -59,7 +64,7 @@ def validate_code_and_proj_string(self) -> "CRS":
                 msg = "code and proj_string must represent the same CRS."
                 raise ValueError(msg)
         return self
-    
+
     def to_string(self) -> str:
         if self.code:
             return self.code
@@ -69,8 +74,8 @@ def to_pyproj(self) -> pyproj.CRS:
         if self.code:
             return pyproj.CRS.from_user_input(self.code)
         return pyproj.CRS.from_user_input(self.proj_string)
-    
-    def __hash__(self):
+
+    def __hash__(self) -> int:
         return hash(self.name)