Derive growth percentile records

1library(yamaa)23advs <- yamaa_domain("spec.yaml", environment = "r/environment.yaml")$output4advs

1import yamaa23advs = yamaa.domain("spec.yaml", environment="python/environment.yaml").output4advs

1# Reference solution for the yamaa benchmark adam-advs-percentiles (R track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67library(dplyr, warn.conflicts = FALSE)8library(readr)910raw <- read_csv(11  "/app/input/advs_raw.csv",12  col_types = cols(AVAL = col_double(), AGE = col_integer(), .default = col_character())13)14ref <- read_csv(15  "/app/input/lms_ref.csv",16  col_types = cols(17    AGE = col_integer(), L = col_double(), M = col_double(), S = col_double(),18    .default = col_character()19  )20)2122# The growth percentile against the matched L, M, and S coefficients.23advs <- raw |>24  left_join(ref, by = c("PARAMCD", "SEX", "AGE")) |>25  mutate(26    Z = case_when(27      is.na(AVAL) | is.na(L) | is.na(M) | is.na(S) | M == 0 | S == 0 ~ NA_real_,28      L == 0 ~ log(AVAL / M) / S,29      .default = ((AVAL / M) ^ L - 1) / (L * S)30    ),31    AVAL = 100 * pnorm(Z),32    PARAMCD = if_else(PARAMCD == "BMI", "BMIPCTL", "WGTPCTL"),33    PARAM = if_else(PARAMCD == "BMIPCTL", "BMI-for-Age Percentile", "Weight-for-Age Percentile")34  ) |>35  arrange(USUBJID, AVISIT, PARAMCD) |>36  select(STUDYID, USUBJID, AVISIT, PARAMCD, PARAM, AVAL)3738dir.create("/app/output", showWarnings = FALSE)39write_csv(advs, "/app/output/advs.csv", na = "")

1# Reference solution for the yamaa benchmark adam-advs-percentiles (Python track).2#3# Written from the benchmark's full prompt and its inputs alone, as an agent4# would write it. Harbor's oracle agent runs it to check that the prompt can5# be solved and that the grader scores a correct answer 1.67import math8from pathlib import Path910import polars as pl1112raw = pl.read_csv("/app/input/advs_raw.csv", infer_schema=False).with_columns(13    pl.col("AVAL").cast(pl.Float64, strict=False),14    pl.col("AGE").cast(pl.Int64, strict=False),15)16ref = pl.read_csv("/app/input/lms_ref.csv", infer_schema=False).with_columns(17    pl.col("AGE").cast(pl.Int64, strict=False),18    pl.col("L").cast(pl.Float64, strict=False),19    pl.col("M").cast(pl.Float64, strict=False),20    pl.col("S").cast(pl.Float64, strict=False),21)222324def percentile(aval: float | None, ell: float | None, emm: float | None, ess: float | None) -> float | None:25    if aval is None or ell is None or emm is None or ess is None:26        return None27    if emm == 0 or ess == 0:28        return None29    if ell == 0:30        z = math.log(aval / emm) / ess31    else:32        z = ((aval / emm) ** ell - 1) / (ell * ess)33    return 100 * 0.5 * (1 + math.erf(z / math.sqrt(2)))343536joined = raw.join(ref, on=["PARAMCD", "SEX", "AGE"], how="left")3738rows = []39for record in joined.sort(["USUBJID", "AVISIT", "PARAMCD"]).iter_rows(named=True):40    if record["PARAMCD"] == "BMI":41        out_code, out_label = "BMIPCTL", "BMI-for-Age Percentile"42    else:43        out_code, out_label = "WGTPCTL", "Weight-for-Age Percentile"44    rows.append(45        {46            "STUDYID": record["STUDYID"],47            "USUBJID": record["USUBJID"],48            "AVISIT": record["AVISIT"],49            "PARAMCD": out_code,50            "PARAM": out_label,51            "AVAL": percentile(record["AVAL"], record["L"], record["M"], record["S"]),52        }53    )5455advs = pl.DataFrame(56    rows,57    schema={58        "STUDYID": pl.String,59        "USUBJID": pl.String,60        "AVISIT": pl.String,61        "PARAMCD": pl.String,62        "PARAM": pl.String,63        "AVAL": pl.Float64,64    },65)6667Path("/app/output").mkdir(parents=True, exist_ok=True)68advs.write_csv("/app/output/advs.csv")

1/* opensas reference solution for the yamaa benchmark adam-advs-percentiles.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data advs_raw;6  length STUDYID USUBJID AVISIT PARAMCD SEX $1024;7  infile "/app/input/advs_raw.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input STUDYID : $char1024. USUBJID : $char1024. AVISIT : $char1024. PARAMCD : $char1024. AVAL AGE SEX : $char1024.;9run;1011data lms_ref;12  length PARAMCD SEX $1024;13  infile "/app/input/lms_ref.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;14  input PARAMCD : $char1024. SEX : $char1024. AGE L M S;15run;1617proc sql;18  create table joined as select a.*,b.L,b.M,b.S from advs_raw as a left join lms_ref as b on a.PARAMCD=b.PARAMCD and a.SEX=b.SEX and a.AGE=b.AGE;19quit;20data result;21  length PARAM $40;22  set joined;23  value=AVAL; AVAL=.;24  if not missing(value) and not missing(M) and not missing(L) and not missing(S) then do;25    if L=0 then z=log(value/M)/S; else z=((value/M)**L-1)/(L*S);26    AVAL=100*probnorm(z);27  end;28  if PARAMCD='BMI' then do; PARAMCD='BMIPCTL'; PARAM='BMI-for-Age Percentile'; end;29  else if PARAMCD='WEIGHT' then do; PARAMCD='WGTPCTL'; PARAM='Weight-for-Age Percentile'; end;30run;3132proc sql;33  create table final as select STUDYID, USUBJID, AVISIT, PARAMCD, PARAM, AVAL from result;34quit;35proc export data=final outfile="/app/output/advs.csv" dbms=csv replace; run;