ltc-code 0.2.27__tar.gz → 0.2.29__tar.gz

{ltc_code-0.2.27 → ltc_code-0.2.29}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ltc-code
-Version: 0.2.27
+Version: 0.2.29
 Summary: Add your description here
 Requires-Dist: fastexcel>=0.16,<0.20
 Requires-Dist: polars>=1.36.1,<1.42

{ltc_code-0.2.27 → ltc_code-0.2.29}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ltc-code"
-version = "0.2.27"
+version = "0.2.29"
 description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.9"
@@ -17,3 +17,7 @@ ltc-summary-stats = "ltc_code.make_summary_stats_table:main"
 [build-system]
 requires = ["uv_build>=0.11.16,<0.12.0"]
 build-backend = "uv_build"
+
+[tool.uv.build-backend]
+source-exclude = ["**/.DS_Store", "**/__pycache__/**", "**/*.pyc", "src/ltc_code/nsc/input/**", "src/ltc_code/nsc/output/**"]
+wheel-exclude = ["**/.DS_Store", "**/__pycache__/**", "**/*.pyc", "ltc_code/nsc/input/**", "ltc_code/nsc/output/**"]

ltc_code-0.2.29/src/ltc_code/nsc/NEW_CROSSWALK.md

@@ -0,0 +1,24 @@
+# New NSC crosswalk build
+
+Run the new variant explicitly with `python -m ltc_code.nsc.build_nsc_outcomes_new` after providing the student input files. The original build remains available. The new variant writes to `nsc/output/new_crosswalk/` to keep its outputs separate.
+
+The official April 2023 NSC crosswalk is primary. Append mapped legacy codes only when their full lookup key is absent from the official file. Do not fill the 108 present-but-blank official keys using Sarah's mappings. Existing 18 sole-rate candidate choices and 24 Calhoun branch assignments are explicit and documented in the code. Preserve full branch codes and deterministic legacy selection.
+
+Institution type uses IPEDS 2013, then the latest nonmissing directory classification, then three inline exceptions formerly loaded from the manual spreadsheet. Less-than-two-year labels are parsed before broader two-year labels. Treat Missing/not reported as null before choosing the latest directory record. The existing two-year-or-less outcome grouping is retained. The 5–8 tier bucket lookup counts Nevada State once despite its tier 5/7 aliases.
+
+All student outcome formulas, eligibility windows, completion-rate imputation, and the existing within-bucket coarse-rate exercise remain unchanged. The change affects their institutional inputs. No annual enrollment-year classification is introduced.
+
+Expected counts with the audited files: 24,166 keys; 21,315 with UNITID; 21,204 matching a UNITID in the 2013/directory files (99.479%); 7,867 matched institutions with no missing type. Distinct institution types: 3,329 four-year, 2,440 two-year, 2,098 less-than-two-year. These are reference-table counts, not secure-sample match rates or completeness of completion rates.
+
+Required files in `nsc/raw/`:
+- NSC_SCHOOL_CODE_TO_IPEDS_UNIT_ID_XWALK_APR-2023.xlsx (official workbook, NSC_to_IPEDS_UNIT_ID sheet)
+- college_crosswalk.xls (legacy fallback)
+- IPEDS_IC_2013.csv
+- directory.dta
+- CREDENTIAL_LEVEL_LOOKUP_TABLE.xlsx
+- ipeds_data.dta
+- chetty/mrc_table2.dta and chetty/mrc_table11.dta
+
+The old IPEDS_IC_manual.xlsx is not needed by this variant. Student inputs remain input/apps.csv, input/nsc_records_old.dta, and input/nsc_records_new.csv. Do not supply synthetic records for actual analysis. Institutional support files are bundled in the locally built wheel and source distribution but are Git-ignored. GitHub Actions publishes these verified distributions from release assets; it does not rebuild from a checkout that lacks the data. Student inputs are never bundled.
+
+Institution-only audit CSVs are written alongside the outcomes, including mapping choices, appended candidates, missing UNITIDs, and missing classifications. The 99.48% matched subset is an audit denominator, not a filter that drops student records from the build.

{ltc_code-0.2.27 → ltc_code-0.2.29}/src/ltc_code/nsc/build_nsc_outcomes.py

@@ -1198,9 +1198,10 @@ for year in range(1, N_YEARS_OUT + 1):
                     [
                         (
                             pl.col("_overlaps")
-                            * pl.col(f"college_{college_type}")
-                            .fill_null(0)
-                            .cast(pl.Int8)
+                            * (
+                                pl.lit(1) if college_type == "any"
+                                else pl.col(f"college_{college_type}").fill_null(0).cast(pl.Int8)
+                            )
                         )
                         .max()
                         .alias(f"att_{college_type}_{label}")
@@ -1258,9 +1259,10 @@ for age in AGE_ATTENDANCE_RANGE:
             [
                 (
                     pl.col("_overlaps")
-                    * pl.col(f"college_{college_type}")
-                    .fill_null(0)
-                    .cast(pl.Int8)
+                    * (
+                        pl.lit(1) if college_type == "any"
+                        else pl.col(f"college_{college_type}").fill_null(0).cast(pl.Int8)
+                    )
                 )
                 .max()
                 .alias(f"att_{college_type}_{age}")
@@ -1878,6 +1880,60 @@ nsc_outcomes = nsc_outcomes.with_columns(
     .alias("adj_cmp_rate_2yr"),
 )
 
+# Separate the fully coarse prediction by first-institution sector.
+nsc_outcomes = nsc_outcomes.with_columns(
+    pl.when(
+        pl.col("adj_cmp_rate_coarse").is_not_null()
+        & pl.col("college_years_firstinst").is_in([1, 2])
+    )
+    .then(0.0)
+    .otherwise(pl.col("adj_cmp_rate_coarse"))
+    .alias("adj_cmp_rate_coarse_4yr"),
+    pl.when(
+        pl.col("adj_cmp_rate_coarse").is_not_null()
+        & (pl.col("college_years_firstinst") == 4)
+    )
+    .then(0.0)
+    .otherwise(pl.col("adj_cmp_rate_coarse"))
+    .alias("adj_cmp_rate_coarse_2yr"),
+)
+
+# Only tiers 1-8 enter these four-year bucket averages. Other-tier branch
+# mappings can share a UNITID and must not duplicate the institution's cohort.
+ipeds_tiers = (
+    college_ref.select("unitid", "tier")
+    .drop_nulls()
+    .filter(pl.col("tier").is_in(range(1, 9)))
+    .unique()
+)
+ipeds_tier_completion = ipeds_completion_with_sector.join(
+    ipeds_tiers, on="unitid", how="left", validate="1:1"
+).filter(pl.col("college_sector") == "4yr")
+
+# Coarsen one bucket at a time, always starting from the detailed prediction.
+tier_buckets = {"t12": [1, 2], "t34": [3, 4], "t58": [5, 6, 7, 8]}
+for label, tiers in tier_buckets.items():
+    bucket_rate = (
+        ipeds_tier_completion.filter(pl.col("tier").is_in(tiers))
+        .select(
+            pl.col("completers_150pct_ip").sum()
+            / pl.col("cohort_adj_150pct_ip").sum()
+        )
+        .item()
+    )
+    nsc_outcomes = nsc_outcomes.with_columns(
+        pl.when(
+            (pl.col("college_years_firstinst") == 4)
+            & pl.col("tier_firstinst").is_in(tiers)
+            & (pl.col("att_any_byY4") == 1)
+            & pl.col("adj_cmp_rate_4yr").is_not_null()
+        )
+        .then(bucket_rate)
+        .otherwise(pl.col("adj_cmp_rate_4yr"))
+        .alias(f"adj_cmp_rate_4yr_coarse_{label}")
+    )
+    print(f"Four-year IPEDS-cohort-weighted completion rate, {label}: {bucket_rate}")
+
 # Match the available 1098-T calendar years, including the missing 2015 year.
 tax_years = list(range(2011, 2015)) + list(range(2016, 2023))
 nsc_outcomes = nsc_outcomes.with_columns(
@@ -1888,6 +1944,7 @@ nsc_outcomes = nsc_outcomes.with_columns(
         .alias(f"att_any_1098_{age}")
         for age in AGE_ATTENDANCE_RANGE
     ]
+
 )
 
 
@@ -1966,6 +2023,9 @@ keep_columns = [
     "adj_cmp_rate",
     "adj_cmp_rate_2yr",
     "adj_cmp_rate_4yr",
+    "adj_cmp_rate_coarse_2yr",
+    "adj_cmp_rate_coarse_4yr",
+    *[f"adj_cmp_rate_4yr_coarse_{label}" for label in tier_buckets],
     "adj_cmp_rate_coarse",
     "adj_cmp_rate_coarse_sample",
     "adj_cmp_rate_coarsen_4yr",
@@ -2046,15 +2106,8 @@ coarse_audit = observable_y4.select(
 
 if coarse_audit.item(0, "k_mean_coarse_missing") > 0:
     raise ValueError("k_mean_coarse is unexpectedly missing for observable students.")
-if coarse_audit.item(0, "adj_cmp_rate_coarse_missing") > 0:
-    raise ValueError(
-        "adj_cmp_rate_coarse is unexpectedly missing for observable students."
-    )
-if coarse_audit.item(0, "adj_cmp_rate_coarse_sample_missing") > 0:
-    raise ValueError(
-        "adj_cmp_rate_coarse_sample is unexpectedly missing for observable students."
-    )
-
+# Attendees without a classified first college can have missing coarse rates.
+# Report those counts below rather than treating them as a build failure.
 print("Coarse outcome audit:")
 print(coarse_audit)