omnidiag / scripts /data /make_sample_batches.py
yahyoha's picture
Claude Opus 5.5
Release: retire BRFSS diabetes (source deploy/v2-platform 8bbbfe9)
c816150
Raw History Blame Contribute Delete
4.07 kB
"""Generate demo CSVs for the batch-prediction page (synthetic, seeded)."""
import csv, random
from pathlib import Path
OUT = Path(__file__).resolve().parents[2] / "data" / "sample_batch"
rng = random.Random(42)
def clip(x, lo, hi):
return max(lo, min(hi, x))
def heart_row():
age = int(clip(rng.gauss(55, 10), 29, 80))
sex = rng.choices(["M", "F"], [0.7, 0.3])[0]
sick = rng.random() < 0.45
cp = rng.choices(["ASY", "NAP", "ATA", "TA"], [0.6, 0.15, 0.15, 0.1] if sick else [0.2, 0.3, 0.4, 0.1])[0]
return {
"Age": age, "Sex": sex, "ChestPainType": cp,
"RestingBP": int(clip(rng.gauss(132, 17), 90, 200)),
"Cholesterol": int(clip(rng.gauss(245, 50), 130, 500)),
"FastingBS": int(rng.random() < (0.25 if sick else 0.1)),
"RestingECG": rng.choices(["Normal", "ST", "LVH"], [0.6, 0.2, 0.2])[0],
"MaxHR": int(clip(rng.gauss(130 if sick else 152, 20), 70, 200)),
"ExerciseAngina": "Y" if rng.random() < (0.55 if sick else 0.12) else "N",
"Oldpeak": round(clip(rng.gauss(1.6 if sick else 0.5, 1.0), 0, 6), 1),
"ST_Slope": rng.choices(["Flat", "Down", "Up"], [0.6, 0.2, 0.2] if sick else [0.2, 0.05, 0.75])[0],
}
def nhanes_row():
"""NHANES dysglycaemia module (configs/diabetes_nhanes.yaml): values inside
the schema bounds, loosely correlated so both decisions occur."""
hi = rng.random() < 0.45
age = int(clip(rng.gauss(58 if hi else 40, 12), 20, 80))
bmi = round(clip(rng.gauss(32 if hi else 25, 5), 16, 60), 1)
band = "high" if bmi >= 32 else "increased" if bmi >= 26 else "normal"
return {
"RIDAGEYR": age, "RIAGENDR": rng.randint(0, 1), "BMXBMI": bmi, "ADIPOSITY_BAND": band,
"SBP": int(clip(rng.gauss(138 if hi else 118, 15), 90, 200)),
"DBP": int(clip(rng.gauss(82 if hi else 72, 10), 50, 120)),
"BPXPLS": int(clip(rng.gauss(74, 10), 45, 120)),
"MCQ300C": int(rng.random() < (0.45 if hi else 0.2)),
"CVD_ANY": int(rng.random() < (0.15 if hi else 0.03)),
"PAQ650": int(rng.random() < (0.15 if hi else 0.35)),
"PAQ665": int(rng.random() < (0.35 if hi else 0.6)),
"LBDHDD": int(clip(rng.gauss(42 if hi else 56, 12), 20, 110)),
"LBXSCH": int(clip(rng.gauss(200, 38), 110, 340)),
"LBXSTR": int(clip(rng.gauss(190 if hi else 110, 60), 40, 600)),
"LBXSATSI": int(clip(rng.gauss(30 if hi else 22, 10), 8, 120)),
"LBXSGTSI": int(clip(rng.gauss(40 if hi else 24, 18), 8, 200)),
"LBXSCR": round(clip(rng.gauss(0.9, 0.2), 0.4, 2.0), 2),
"LBXSBU": int(clip(rng.gauss(14, 4), 5, 40)),
"LBXSAL": round(clip(rng.gauss(4.3, 0.3), 3.2, 5.2), 1),
"LBXSUA": round(clip(rng.gauss(6.2 if hi else 5.2, 1.2), 2.5, 10.0), 1),
}
def write(name, rows):
with open(OUT / name, "w", newline="", encoding="utf-8") as f:
w = csv.DictWriter(f, fieldnames=list(rows[0].keys()))
w.writeheader()
w.writerows(rows)
# Heart: 100 rows; 6 with missing optional values (valid), 3 invalid (error rows)
heart = [heart_row() for _ in range(100)]
for i in (4, 17, 33, 58, 71, 90):
heart[i]["Cholesterol"] = ""
if i % 2:
heart[i]["Oldpeak"] = ""
heart[10]["Age"] = 150 # out of range
heart[45]["ChestPainType"] = "XYZ" # invalid category
heart[80]["Sex"] = "" # required field missing
write("heart_disease_batch_sample.csv", heart)
# NHANES dysglycaemia: the module cap (max_batch_rows) is 100 rows. 6 with
# missing optional values (valid), 3 invalid (error rows). The BRFSS sample
# (diabetes_batch_sample.csv) went with that module in gate B7.
nhanes = [nhanes_row() for _ in range(100)]
for i in (3, 21, 39, 52, 68, 84):
nhanes[i]["LBXSTR"] = ""
if i % 2:
nhanes[i]["SBP"] = nhanes[i]["DBP"] = ""
nhanes[12]["RIDAGEYR"] = 15 # below the adult cohort (20-80)
nhanes[47]["ADIPOSITY_BAND"] = "XL" # invalid category
nhanes[77]["LBDHDD"] = "" # required field missing
write("diabetes_nhanes_batch_sample.csv", nhanes)
print("ok")