publicdata/j40-cejst-2
1
0
Fork 0
mirror of https://github.com/DOI-DO/j40-cejst-2.git synced 2025-02-23 01:54:18 -08:00
Code Issues Projects Releases Packages Wiki Activity Actions

updated with some matt comments

Browse source
This commit is contained in:
Emma Nechamkin 2022-09-07 15:22:13 -04:00
parent 6d9e11d081
commit 0115239e50
1 changed files with 25 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

45
data/data-pipeline/data_pipeline/etl/sources/census_acs/etl_imputations.py
View file

@ -10,11 +10,11 @@ from data_pipeline.utils import get_module_logger
logger = get_module_logger(__name__)
def _prepare_dataframe_for_imputation(
def _get_impute_tract_list(
columns_to_impute: list,
geo_df: gpd.GeoDataFrame,
geoid_field: str = "GEOID10_TRACT",
):
) -> list:
# generate a list of tracts for which at least one of the imputation
# columns is null
@ -26,30 +26,35 @@ def _prepare_dataframe_for_imputation(
logger.info(f"Imputing values for {len(tract_list)} unique tracts.")
assert len(tract_list) > 0, "Error: No missing values to impute"
return tract_list, geo_df
return tract_list
def _get_state_and_county_fills(df, tract_list, impute_var_pair_list):
def _get_state_and_county_fills(
df: pd.DataFrame, tract_list: list, impute_var_pair_list: list
) -> pd.DataFrame:
counties = df[GEOID_TRACT_FIELD].str[:5] # county fips is 5 digits
states = df[GEOID_TRACT_FIELD].str[:2] # state fips is 2 digits
# When there is no neighbor average, we take the county-level average or state-level averages
for impute_var_pair in impute_var_pair_list:
# Fill missings with county means
df[impute_var_pair.imputed_field_name] = np.where(
(df[impute_var_pair.imputed_field_name].isna())
& (df[GEOID_TRACT_FIELD].isin(tract_list)),
df.groupby(df[GEOID_TRACT_FIELD].str[:5])[
# Get a column of county means or state means when county means are not available
county_means = df.groupby(counties)[
impute_var_pair.raw_field_name
].transform(np.mean),
df[impute_var_pair.imputed_field_name],
)
# Fill the remaining missings with state means
df[impute_var_pair.imputed_field_name] = np.where(
(df[impute_var_pair.imputed_field_name].isna())
& (df[GEOID_TRACT_FIELD].isin(tract_list)),
df.groupby(df[GEOID_TRACT_FIELD].str[:2])[
].transform(np.mean)
state_means = df.groupby(states)[
impute_var_pair.raw_field_name
].transform(np.mean),
df[impute_var_pair.imputed_field_name],
].transform(np.mean)
fill_means = county_means.fillna(state_means)
# Identify where these must be imputed
impute_tracts = (df[impute_var_pair.imputed_field_name].isna()) & (
df[GEOID_TRACT_FIELD].isin(tract_list)
)
# And then impute while preserving null character elsewhere
df[impute_var_pair.imputed_field_name] = np.where(
impute_tracts, fill_means, df[impute_var_pair.imputed_field_name]
)
return df
@ -82,7 +87,7 @@ def calculate_income_measures(
rename_dict[impute_var.raw_field_name] = impute_var.imputed_field_name
# Determine where to impute variables and fill a column with nulls
tract_list, geo_df = _prepare_dataframe_for_imputation(
tract_list = _get_impute_tract_list(
columns_to_impute=raw_fields,
geo_df=geo_df,
geoid_field=geoid_field,