Add cleaning script for the generated health module (SF-12)

src/soep_preparation/clean_modules/health.py

@@ -0,0 +1,75 @@
+"""Clean and convert SOEP health variables to appropriate data types."""
+
+import pandas as pd
+
+from soep_preparation.utilities.data_manipulator import (
+    apply_smallest_int_dtype,
+    create_dummy,
+    float_to_float,
+)
+
+
+def clean(raw_data: pd.DataFrame) -> pd.DataFrame:
+    """Create cleaned variables from the health module.
+
+    The health module is a generated person-year dataset with health
+    indicators collected in a two-year replication cycle since 2002. Its
+    centerpiece is the SOEP version of the SF-12v2: eight norm-based
+    subscales plus the physical (PCS) and mental (MCS) component summary
+    scales, all standardized to mean 50 and standard deviation 10 in the
+    SOEP 2004 population, with higher values representing better
+    health-related quality of life.
+
+    Args:
+        raw_data: The raw health data.
+
+    Returns:
+        The processed health data.
+    """
+    out = pd.DataFrame()
+
+    out["p_id"] = apply_smallest_int_dtype(raw_data["pid"])
+    out["hh_id_original"] = apply_smallest_int_dtype(raw_data["cid"])
+    out["survey_year"] = apply_smallest_int_dtype(raw_data["syear"])
+
+    # Component summary scales. The SF-12 scores and BMI arrive as floats with
+    # SOEP missing codes encoded as negative values.
+    out["sf12_pcs"] = float_to_float(raw_data["pcs"], code_negative_values_as_na=True)
+    out["sf12_mcs"] = float_to_float(raw_data["mcs"], code_negative_values_as_na=True)
+
+    # The eight norm-based subscales underlying the summary scales.
+    out["sf12_physical_functioning_nbs"] = float_to_float(
+        raw_data["pf_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_role_physical_nbs"] = float_to_float(
+        raw_data["rp_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_bodily_pain_nbs"] = float_to_float(
+        raw_data["bp_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_general_health_nbs"] = float_to_float(
+        raw_data["gh_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_vitality_nbs"] = float_to_float(
+        raw_data["vt_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_social_functioning_nbs"] = float_to_float(
+        raw_data["sf_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_role_emotional_nbs"] = float_to_float(
+        raw_data["re_nbs"], code_negative_values_as_na=True
+    )
+    out["sf12_mental_health_nbs"] = float_to_float(
+        raw_data["mh_nbs"], code_negative_values_as_na=True
+    )
+
+    # Whether all twelve items needed for the SF-12 scoring are complete.
+    out["sf12_valid"] = create_dummy(
+        series=raw_data["valid"],
+        value_for_comparison="[1] Yes",
+        comparison_type="equal",
+    )
+
+    out["bmi_health"] = float_to_float(raw_data["bmi"], code_negative_values_as_na=True)
+
+    return out