{
  "alt": "Dot chart of three data-chosen cutoffs on a probability scale with a dashed rule at the cost cutoff p*. Jev's cutoff is furthest right, Luna's is between, and the classifier's sits next to the rule.",
  "caption": "The dashed rule is the cost cutoff p* = {fact:p_star}. Dots to its right belong to models whose probabilities overstate dissatisfaction near the cutoff, so the data push the cutoff up to contact fewer satisfied reviewers.",
  "data": [
    {
      "cutoff": 0.38,
      "engine": "Jev 1.14.0",
      "fold_high": 0.38,
      "fold_low": 0.35
    },
    {
      "cutoff": 0.22,
      "engine": "OpenAI Luna",
      "fold_high": 0.29,
      "fold_low": 0.2
    },
    {
      "cutoff": 0.114,
      "engine": "TF-IDF + logistic",
      "fold_high": 0.156,
      "fold_low": 0.09
    }
  ],
  "id": "data-chosen-cutoffs",
  "notes": [
    "Each dot is the median of the five fold cutoffs; the table gives the lowest and highest. No bootstrap interval was computed."
  ],
  "source": "Amazon Reviews 2023, coffee products; 5,000 sampled reviews, weighted; engine calls October 2026; assumed costs",
  "spec": {
    "reference": 0.125,
    "referenceLabel": "Cost cutoff p*",
    "sort": "data",
    "valueLabels": true,
    "x": "cutoff",
    "xDomain": [
      0,
      0.5
    ],
    "xFormat": {
      "digits": 2
    },
    "xLabel": "Cutoff on the probability of dissatisfaction",
    "y": "engine"
  },
  "subtitle": "Cutoff that maximizes net value, chosen on four folds of the sample and applied to the fifth, median of five folds, by model",
  "table": {
    "columns": [
      {
        "key": "engine",
        "label": "Model"
      },
      {
        "format": {
          "digits": 2
        },
        "key": "cutoff",
        "label": "Median cutoff"
      },
      {
        "format": {
          "digits": 2
        },
        "key": "fold_low",
        "label": "Lowest fold"
      },
      {
        "format": {
          "digits": 2
        },
        "key": "fold_high",
        "label": "Highest fold"
      }
    ]
  },
  "title": "Letting the data choose the cutoff moves it above p* for the engines",
  "type": "dot",
  "width": "body"
}