Carry one observed record to each planned visit

1library(yamaa)23advs <- yamaa_domain("spec.yaml")$output4advs

1import yamaa23advs = yamaa.yamaa_domain("spec.yaml").output4advs

1# Reference solution for the yamaa benchmark adam-advs-locf-record (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(purrr)9library(readr)10library(tidyr)1112plan <- read_csv(13  "/app/input/plan.csv",14  col_types = cols(AVISITN = col_integer(), .default = col_character())15)16vs <- read_csv(17  "/app/input/vs.csv",18  col_types = cols(19    AVISITN = col_integer(), AVAL = col_double(), ADT = col_date(),20    QSSEQ = col_integer(), .default = col_character()21  )22)2324# Only results selected for analysis can be carried.25candidates <- vs |>26  filter(ANL01FL %in% "Y", !is.na(AVAL))2728# The latest non-missing selected result at or before the planned visit:29# the highest visit number, then the highest sequence number.30pick_donor <- function(usubjid, paramcd, avisitn) {31  donor <- candidates |>32    filter(USUBJID == usubjid, PARAMCD == paramcd, AVISITN <= avisitn) |>33    arrange(AVISITN, QSSEQ) |>34    slice_tail(n = 1)35  if (nrow(donor) == 0) {36    tibble::tibble(AVAL = NA_real_, ADT = as.Date(NA), QSSEQ = NA_integer_)37  } else {38    donor |> select(AVAL, ADT, QSSEQ)39  }40}4142advs <- plan |>43  arrange(USUBJID, PARAMCD, AVISITN) |>44  mutate(donor = pmap(list(USUBJID, PARAMCD, AVISITN), pick_donor)) |>45  unnest(donor) |>46  select(USUBJID, PARAMCD, AVISITN, AVAL, ADT, QSSEQ)4748dir.create("/app/output", showWarnings = FALSE)49write_csv(advs, "/app/output/advs.csv", na = "")

1# Reference solution for the yamaa benchmark adam-advs-locf-record (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 1.67from pathlib import Path89import polars as pl1011plan = pl.read_csv("/app/input/plan.csv", infer_schema=False).with_columns(12    pl.col("AVISITN").cast(pl.Int64, strict=False)13)14vs = pl.read_csv("/app/input/vs.csv", infer_schema=False).with_columns(15    pl.col("AVISITN").cast(pl.Int64, strict=False),16    pl.col("AVAL").cast(pl.Float64, strict=False),17    pl.col("ADT").str.to_date(strict=False),18    pl.col("QSSEQ").cast(pl.Int64, strict=False),19)2021# Only results selected for analysis can be carried.22candidates = vs.filter(23    (pl.col("ANL01FL") == "Y") & pl.col("AVAL").is_not_null()24)2526# The latest non-missing selected result at or before the planned visit:27# the highest visit number, then the highest sequence number.28rows = []29for planned in plan.sort(["USUBJID", "PARAMCD", "AVISITN"]).iter_rows(named=True):30    donor = (31        candidates.filter(32            (pl.col("USUBJID") == planned["USUBJID"])33            & (pl.col("PARAMCD") == planned["PARAMCD"])34            & (pl.col("AVISITN") <= planned["AVISITN"])35        )36        .sort(["AVISITN", "QSSEQ"])37        .tail(1)38    )39    if donor.height:40        record = donor.row(0, named=True)41        rows.append(42            {43                "USUBJID": planned["USUBJID"],44                "PARAMCD": planned["PARAMCD"],45                "AVISITN": planned["AVISITN"],46                "AVAL": record["AVAL"],47                "ADT": record["ADT"],48                "QSSEQ": record["QSSEQ"],49            }50        )51    else:52        rows.append(53            {54                "USUBJID": planned["USUBJID"],55                "PARAMCD": planned["PARAMCD"],56                "AVISITN": planned["AVISITN"],57                "AVAL": None,58                "ADT": None,59                "QSSEQ": None,60            }61        )6263advs = pl.DataFrame(64    rows,65    schema={66        "USUBJID": pl.String,67        "PARAMCD": pl.String,68        "AVISITN": pl.Int64,69        "AVAL": pl.Float64,70        "ADT": pl.Date,71        "QSSEQ": pl.Int64,72    },73).sort(["USUBJID", "PARAMCD", "AVISITN"])7475Path("/app/output").mkdir(parents=True, exist_ok=True)76advs.write_csv("/app/output/advs.csv")

1/* opensas reference solution for the yamaa benchmark adam-advs-locf-record.2   Reads the benchmark inputs and derives the requested output.3   Execute with the existing opensas runtime: opensas /app/output/result.sas. */45data plan;6  length USUBJID PARAMCD $1024;7  infile "/app/input/plan.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;8  input USUBJID : $char1024. PARAMCD : $char1024. AVISITN;9run;1011data vs;12  length USUBJID PARAMCD ADT ANL01FL $1024;13  infile "/app/input/vs.csv" dsd dlm="," firstobs=2 truncover lrecl=32767;14  input USUBJID : $char1024. PARAMCD : $char1024. AVISITN AVAL ADT : $char1024. QSSEQ ANL01FL : $char1024.;15run;1617proc sql;18  create table candidates as select a.USUBJID,a.PARAMCD,a.AVISITN,b.AVAL,b.ADT,b.QSSEQ,b.AVISITN as observedvisit from plan as a left join vs as b19    on a.USUBJID=b.USUBJID and a.PARAMCD=b.PARAMCD and b.AVISITN<=a.AVISITN and b.ANL01FL='Y' and not missing(b.AVAL);20quit;21proc sort data=candidates; by USUBJID PARAMCD AVISITN descending observedvisit descending QSSEQ; run;22data result; set candidates; by USUBJID PARAMCD AVISITN; if first.AVISITN; run;2324proc sql;25  create table final as select USUBJID, PARAMCD, AVISITN, AVAL, ADT, QSSEQ from result;26quit;27proc export data=final outfile="/app/output/advs.csv" dbms=csv replace; run;