{
  "alt": "Dumbbell chart with rows for Jev, Luna and the TF-IDF classifier. Jev's filled dot sits clearly right of its open dot, Luna's slightly right, and the classifier's slightly left.",
  "caption": "The open dot is net value at p*; the filled dot is net value at the cutoff the data chose, scored on held-out folds. The last column is the gain. Whiskers on the open dots are 95% intervals from a bootstrap that resamples products.",
  "data": [
    {
      "at_data": 5335,
      "at_formula": 5076,
      "engine": "Jev 1.14.0",
      "formula_hi": 5144,
      "formula_lo": 5004,
      "n": 5000
    },
    {
      "at_data": 5273,
      "at_formula": 5195,
      "engine": "OpenAI Luna",
      "formula_hi": 5253,
      "formula_lo": 5135,
      "n": 5000
    },
    {
      "at_data": 4985,
      "at_formula": 5015,
      "engine": "TF-IDF + logistic",
      "formula_hi": 5104,
      "formula_lo": 4928,
      "n": 5000
    }
  ],
  "id": "net-at-data-cutoffs",
  "notes": [
    "The data-chosen value is scored out of fold and carries no bootstrap interval."
  ],
  "source": "Amazon Reviews 2023, coffee products; 5,000 sampled reviews, weighted; engine calls October 2026; assumed costs",
  "spec": {
    "deltaLabel": "Gain (US dollars)",
    "hi1": "formula_hi",
    "label1": "At p*",
    "label2": "At the data-chosen cutoff",
    "lo1": "formula_lo",
    "n1": "n",
    "n2": "n",
    "sort": "data",
    "x1": "at_formula",
    "x2": "at_data",
    "xDomain": [
      4500,
      5600
    ],
    "xFormat": {
      "digits": 0,
      "prefix": "$",
      "thousands": true
    },
    "xLabel": "Net value (US dollars per thousand reviews)",
    "y": "engine"
  },
  "subtitle": "Net value per thousand reviews at the cost cutoff p* and at the data-chosen cutoff, by model, weighted",
  "title": "A data-chosen cutoff adds the most net value for Jev, the least calibrated engine, and nothing for the classifier",
  "type": "dumbbell",
  "width": "body"
}