Merge pull request #251 from joshuacortez/chore/testing_notebooks

Chore / Making notebooks up-to-date
thinkingmachines · Aug 14, 2024 · caacdd3 · caacdd3
2 parents a6abea3 + 39ed72b
commit caacdd3
Show file tree

Hide file tree

Showing 23 changed files with 1,127 additions and 1,509 deletions.
diff --git a/geowrangler/area_zonal_stats.py b/geowrangler/area_zonal_stats.py
@@ -3,15 +3,15 @@
 # %% auto 0
 __all__ = ['create_area_zonal_stats']
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 9
+# %% ../notebooks/06_area_zonal_stats.ipynb 7
 from typing import Any, Dict, List
 
 import geopandas as gpd
 import numpy as np
 import geowrangler.vector_zonal_stats as vzs
 from .vector_zonal_stats import GEO_INDEX_NAME
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 11
+# %% ../notebooks/06_area_zonal_stats.ipynb 9
 def extract_func(func):
     # extra by default is none
     extra = []
@@ -32,7 +32,7 @@ def extract_func(func):
 
     return func, extra
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 13
+# %% ../notebooks/06_area_zonal_stats.ipynb 11
 def fix_area_agg(agg):
     if "func" not in agg:
         return agg  # skip fix as agg spec is invalid
@@ -53,7 +53,7 @@ def fix_area_agg(agg):
 
     return agg
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 16
+# %% ../notebooks/06_area_zonal_stats.ipynb 14
 def get_source_column(agg):
 
     if "raw" in agg["extras"]:
@@ -74,35 +74,35 @@ def get_source_column(agg):
         return intersect_aoi_column
     return agg["column"]  # everything else based on raw column
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 18
+# %% ../notebooks/06_area_zonal_stats.ipynb 16
 INTERSECT_AREA_AGG = {
     "column": "intersect_area",
     "func": "sum",
     "output": "intersect_area_sum",
     "extras": "raw",
 }
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 19
+# %% ../notebooks/06_area_zonal_stats.ipynb 17
 def build_agg_area_dicts(aggs):
     aggs = [INTERSECT_AREA_AGG, *aggs]
     agg_dicts = {agg["output"]: (get_source_column(agg), agg["func"]) for agg in aggs}
     return agg_dicts
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 22
+# %% ../notebooks/06_area_zonal_stats.ipynb 20
 def validate_area_aoi(aoi):
     if aoi.crs.is_geographic:
         raise ValueError(
             f"aoi has geographic crs: {aoi.crs}, areas maybe incorrectly computed"
         )
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 23
+# %% ../notebooks/06_area_zonal_stats.ipynb 21
 def validate_area_data(data):
     if data.crs.is_geographic:
         raise ValueError(
             f"data has geographic crs: {data.crs}, areas maybe incorrectly computed"
         )
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 24
+# %% ../notebooks/06_area_zonal_stats.ipynb 22
 def expand_area_aggs(aggs):
     expanded_aggs = []
     for agg in aggs:
@@ -117,7 +117,7 @@ def expand_area_aggs(aggs):
             expanded_aggs += [expanded_agg]
     return expanded_aggs
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 25
+# %% ../notebooks/06_area_zonal_stats.ipynb 23
 def compute_intersect_stats(intersect, expanded_aggs):
     # optimization - use df.apply to create all new columns simultaneously
     for agg in expanded_aggs:
@@ -135,7 +135,7 @@ def compute_intersect_stats(intersect, expanded_aggs):
             )
     return intersect
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 26
+# %% ../notebooks/06_area_zonal_stats.ipynb 24
 def compute_imputed_stats(results, expanded_aggs):
     # optimize with df.apply
     # handle when intersect_area_sum is np.nan
@@ -147,7 +147,7 @@ def compute_imputed_stats(results, expanded_aggs):
 
     return results
 
-# %% ../notebooks/06_area_zonal_stats.ipynb 27
+# %% ../notebooks/06_area_zonal_stats.ipynb 25
 def create_area_zonal_stats(
     aoi: gpd.GeoDataFrame,  # Area of interest for which zonal stats are to be computed for
     data: gpd.GeoDataFrame,  # Source gdf of region/areas containing data to compute zonal stats from

diff --git a/geowrangler/distance_zonal_stats.py b/geowrangler/distance_zonal_stats.py
@@ -3,7 +3,7 @@
 # %% auto 0
 __all__ = ['create_distance_zonal_stats']
 
-# %% ../notebooks/07_distance_zonal_stats.ipynb 9
+# %% ../notebooks/07_distance_zonal_stats.ipynb 7
 from typing import Any, Dict, List
 
 import geopandas as gpd
@@ -12,7 +12,7 @@
 import geowrangler.vector_zonal_stats as vzs
 from .vector_zonal_stats import GEO_INDEX_NAME
 
-# %% ../notebooks/07_distance_zonal_stats.ipynb 11
+# %% ../notebooks/07_distance_zonal_stats.ipynb 9
 INTERNAL_DISTANCE_COL = "__Geowrangleer_nearest_distance"
 
 
@@ -22,7 +22,7 @@ def build_agg_distance_dicts(expanded_aggs, distance_col):
         agg_dicts[distance_col] = (INTERNAL_DISTANCE_COL, "mean")
     return agg_dicts
 
-# %% ../notebooks/07_distance_zonal_stats.ipynb 12
+# %% ../notebooks/07_distance_zonal_stats.ipynb 10
 def create_distance_zonal_stats(
     aoi: gpd.GeoDataFrame,  # Area of interest for which zonal stats are to be computed for
     data: gpd.GeoDataFrame,  # Source gdf of region/areas containing data to compute zonal stats from

diff --git a/geowrangler/raster_to_dataframe.py b/geowrangler/raster_to_dataframe.py
@@ -3,7 +3,7 @@
 # %% auto 0
 __all__ = ['GRID_ID', 'generate_mask', 'read_bands']
 
-# %% ../notebooks/08_raster_to_dataframe.ipynb 10
+# %% ../notebooks/08_raster_to_dataframe.ipynb 8
 import json
 from typing import Any, Dict, List
 
@@ -13,7 +13,7 @@
 import rasterio as rio
 import rasterio.mask
 
-# %% ../notebooks/08_raster_to_dataframe.ipynb 13
+# %% ../notebooks/08_raster_to_dataframe.ipynb 11
 def _explode(gdf):
     """
     Explodes a geodataframe
@@ -38,7 +38,7 @@ def _explode(gdf):
 
     return gdf_out
 
-# %% ../notebooks/08_raster_to_dataframe.ipynb 14
+# %% ../notebooks/08_raster_to_dataframe.ipynb 12
 GRID_ID = 1
 
 
@@ -116,7 +116,7 @@ def generate_mask(
 
     return masks, grids, label_values
 
-# %% ../notebooks/08_raster_to_dataframe.ipynb 16
+# %% ../notebooks/08_raster_to_dataframe.ipynb 14
 def read_bands(image_list: List[str], mask: str):
 
     data = []

diff --git a/geowrangler/spatialjoin_highest_intersection.py b/geowrangler/spatialjoin_highest_intersection.py
@@ -3,15 +3,15 @@
 # %% auto 0
 __all__ = ['get_highest_intersection']
 
-# %% ../notebooks/12_spatialjoin_highest_intersection.ipynb 8
+# %% ../notebooks/12_spatialjoin_highest_intersection.ipynb 6
 import json
 import os
 import geopandas as gpd
 import pandas as pd
 import requests
 from . import grids
 
-# %% ../notebooks/12_spatialjoin_highest_intersection.ipynb 28
+# %% ../notebooks/12_spatialjoin_highest_intersection.ipynb 23
 def get_highest_intersection(
     gdf1: gpd.GeoDataFrame,  # gdf1 will be the basis of output geometry
     gdf2: gpd.GeoDataFrame,  # gdf2 data will all be included during intersection

diff --git a/geowrangler/vector_to_raster_mask.py b/geowrangler/vector_to_raster_mask.py
@@ -3,7 +3,7 @@
 # %% auto 0
 __all__ = ['GRID_ID', 'generate_mask']
 
-# %% ../notebooks/10_vector_to_raster_mask.ipynb 10
+# %% ../notebooks/10_vector_to_raster_mask.ipynb 8
 import json
 from typing import Any, Dict
 
@@ -13,7 +13,7 @@
 import rasterio as rio
 import rasterio.mask
 
-# %% ../notebooks/10_vector_to_raster_mask.ipynb 12
+# %% ../notebooks/10_vector_to_raster_mask.ipynb 10
 def _explode(gdf):
     """
     Explodes a geodataframe
@@ -38,7 +38,7 @@ def _explode(gdf):
 
     return gdf_out
 
-# %% ../notebooks/10_vector_to_raster_mask.ipynb 13
+# %% ../notebooks/10_vector_to_raster_mask.ipynb 11
 GRID_ID = 1
 
 

diff --git a/geowrangler/vector_zonal_stats.py b/geowrangler/vector_zonal_stats.py
@@ -3,10 +3,10 @@
 # %% auto 0
 __all__ = ['create_zonal_stats', 'compute_quadkey', 'create_bingtile_zonal_stats']
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 8
+# %% ../notebooks/02_vector_zonal_stats.ipynb 6
 GEO_INDEX_NAME = "__GeoWrangleer_aoi_index"
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 9
+# %% ../notebooks/02_vector_zonal_stats.ipynb 7
 from functools import partial
 from typing import Any, Dict, List
 
@@ -15,7 +15,7 @@
 import numpy as np
 import pandas as pd
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 13
+# %% ../notebooks/02_vector_zonal_stats.ipynb 11
 def _fix_agg(
     agg: Dict[str, Any],  # A dict containing at the minimum a 'func' key
 ) -> Dict[str, Any]:
@@ -54,7 +54,7 @@ def _fix_agg(
 
     return agg
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 18
+# %% ../notebooks/02_vector_zonal_stats.ipynb 16
 def _check_agg(
     agg: Dict[str, Any],  # A dict containing at the minimum a 'func' key
     i: int,  # The index into the list of aggregations
@@ -95,7 +95,7 @@ def _check_agg(
             f"fillna list {agg['fillna']} doesn't match func list {agg['func']} in agg[{i}] {agg}"
         )
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 21
+# %% ../notebooks/02_vector_zonal_stats.ipynb 19
 def _validate_aggs(
     fixed_aggs: List[Dict[str, Any]],  # A list of fixed agg specs
     data: pd.DataFrame,  # Source dataframe
@@ -111,7 +111,7 @@ def _validate_aggs(
             )
         outputs += agg["output"]
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 26
+# %% ../notebooks/02_vector_zonal_stats.ipynb 24
 def _validate_aoi(
     aoi: pd.DataFrame,  # Source dataframe
 ) -> None:
@@ -120,7 +120,7 @@ def _validate_aoi(
             "AOI has a pandas.MultiIndex. Please convert the index to a single level such as pd.RangeIndex"
         )
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 27
+# %% ../notebooks/02_vector_zonal_stats.ipynb 25
 def _expand_aggs(
     aggs: List[Dict[str, Any]],  # List of fixed valid aggs
 ) -> List[Dict[str, Any]]:
@@ -137,14 +137,14 @@ def _expand_aggs(
             expanded_aggs += [expanded_agg]
     return expanded_aggs
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 29
+# %% ../notebooks/02_vector_zonal_stats.ipynb 27
 def _build_agg_args(
     aggs: List[Dict[str, Any]],  # A list of expanded aggs
 ) -> Dict:
     """Builds a dict of args with output as key and a tuple of column and func as value from a list of expanded aggs"""
     return {agg["output"]: (agg["column"], agg["func"]) for agg in aggs}
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 31
+# %% ../notebooks/02_vector_zonal_stats.ipynb 29
 def _prep_aoi(
     aoi: pd.DataFrame,  # Area of interest
 ) -> pd.DataFrame:
@@ -164,7 +164,7 @@ def _prep_aoi(
     aoi = aoi.reset_index(level=0)  # index added as new column named GEO_INDEX_NAME
     return aoi
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 36
+# %% ../notebooks/02_vector_zonal_stats.ipynb 34
 def _fillnas(
     expanded_aggs: List[Dict[str, Any]],  # list of expanded aggs
     results: pd.DataFrame,  # results dataframe to be filled with NAs if flag set
@@ -186,7 +186,7 @@ def _fillnas(
 
     return results
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 40
+# %% ../notebooks/02_vector_zonal_stats.ipynb 38
 def _aggregate_stats(
     aoi: pd.DataFrame,  # Area of interest
     groups: pd.core.groupby.DataFrameGroupBy,  # Source data aggregated into groups by GEO_INDEX_NAME
@@ -204,7 +204,7 @@ def _aggregate_stats(
 
     return results
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 46
+# %% ../notebooks/02_vector_zonal_stats.ipynb 44
 def create_zonal_stats(
     aoi: gpd.GeoDataFrame,  # Area of interest for which zonal stats are to be computed for
     data: gpd.GeoDataFrame,  # Source gdf containing data to compute zonal stats from
@@ -248,14 +248,14 @@ def create_zonal_stats(
 
     return results
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 62
+# %% ../notebooks/02_vector_zonal_stats.ipynb 60
 tms = morecantile.tms.get("WebMercatorQuad")  # Tile Matrix for Bing Maps
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 63
+# %% ../notebooks/02_vector_zonal_stats.ipynb 61
 def get_quadkey(geometry, zoom_level):
     return tms.quadkey(tms.tile(geometry.x, geometry.y, zoom_level))
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 64
+# %% ../notebooks/02_vector_zonal_stats.ipynb 62
 def compute_quadkey(
     data: gpd.GeoDataFrame,  # The geodataframe
     zoom_level: int,  # The quadkey zoom level (1-23)
@@ -282,7 +282,7 @@ def compute_quadkey(
 
     return data
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 71
+# %% ../notebooks/02_vector_zonal_stats.ipynb 69
 def validate_aoi_quadkey(aoi, aoi_quadkey_column) -> None:
 
     if aoi_quadkey_column not in list(aoi.columns.values):
@@ -310,7 +310,7 @@ def validate_data_quadkey(data, data_quadkey_column, min_zoom_level):
             f"data quadkey levels cannot be less than aoi quadkey level {min_zoom_level}"
         )
 
-# %% ../notebooks/02_vector_zonal_stats.ipynb 72
+# %% ../notebooks/02_vector_zonal_stats.ipynb 70
 def create_bingtile_zonal_stats(
     aoi: pd.DataFrame,  # An aoi with quadkey column
     data: pd.DataFrame,  # Data with  quadkey column