ChallengesDocs

Q2 Solution

Submit the models every medical center runs, with counts

Q2 Solution

POST /api/v2/q2/solution

Request Body

FieldTypeNotes
modelsarrayEvery pair present at all of A, B, C and D
models[].manufacturerstringManufacturer (MFG) as it appears in the records
models[].modelstringModel Name as it appears in the records
models[].totalintegerRecords with this pair across A to D
models[].per_facilityobjectRecords with this pair at each of A, B, C, D
{
  "models": [
    {
      "manufacturer": "WELCH ALLYN",
      "model": "SURETEMP PLUS 690",
      "total": 150,
      "per_facility": { "A": 111, "B": 1, "C": 35, "D": 3 }
    }
  ]
}

Names are compared the same way as in Q1: ignoring case and runs of whitespace, nothing else. Facility E records don't count anywhere in this question.

Responses

200

{
  "challenge": "q2",
  "score": 81,
  "max_points": 100,
  "best_score": 81,
  "result": {
    "models": { "precision": 0.95, "recall": 0.927, "f1": 0.938 },
    "counts_accuracy": 0.683,
    "extra_models": 2
  },
  "attempts_remaining": 4
}
  • models compares the pairs you listed with the pairs that are really shared.
  • counts_accuracy is the average, over the shared pairs, of how many of the five counts you got exactly right. Pairs you missed count as zero.
  • extra_models is how many pairs you listed that aren't shared by all four medical centers.

The score is 50% models.f1 and 50% counts_accuracy.

400

{
  "error": "INVALID_SUBMISSION",
  "message": "models[2].per_facility.C must be a whole number"
}

Invalid requests don't use a submission.

403

{
  "error": "ATTEMPT_LIMIT",
  "message": "You have used all 5 submissions for q2. Your best score of 81 is final.",
  "best_score": 81
}