{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "1beaa094",
   "metadata": {},
   "source": [
    "# ML vs LLM: shared-metric comparison on CA prediction\n",
    "\n",
    "**Goal.** Evaluate traditional machine-learning models (Random Forest, KNN) against LLM persona agents on the **same** prediction task and the **same** metrics.\n",
    "\n",
    "Both families predict PRCA **group** and **interpersonal** subscale scores (6\u201330) from the curated persona information tiers (`demos` \u2192 `employment` \u2192 `geo` \u2192 `transit`).\n",
    "\n",
    "| Metric family | Indicators |\n",
    "|---|---|\n",
    "| Score precision | MAE, exact integer match |\n",
    "| Band agreement | low / moderate / high accuracy |\n",
    "| Distance from correct | normalized score distance (`|e|/24`), ordinal band distance (0\u20132) |\n",
    "\n",
    "Supporting code: [`src/ca_personas/compare_agents.py`](../src/ca_personas/compare_agents.py)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b3f091c",
   "metadata": {},
   "source": [
    "## 1. Setup\n",
    "\n",
    "Set `LLM_PROVIDER` to `mock` for offline runs, or `ollama` / `openrouter` for live models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "68eedba3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:40.110912Z",
     "iopub.status.busy": "2026-07-25T15:40:40.110801Z",
     "iopub.status.idle": "2026-07-25T15:40:41.071630Z",
     "shell.execute_reply": "2026-07-25T15:40:41.070783Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data source: /tmp/sibling_data File A/B/C\n",
      "Prolific: ['/tmp/sibling_data/PRCAProlificExport_FileA.csv', '/tmp/sibling_data/PRCAProlificExport_FileB.csv']\n",
      "Qualtrics: /tmp/sibling_data/PRCAQualtricsExport_FileC.csv\n",
      "LLM provider: mock | tiers: ['demos', 'employment', 'geo', 'transit']\n"
     ]
    }
   ],
   "source": [
    "from __future__ import annotations\n",
    "\n",
    "import os\n",
    "import sys\n",
    "from pathlib import Path\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "\n",
    "ROOT = Path.cwd()\n",
    "if not (ROOT / \"src\").exists() and (ROOT.parent / \"src\").exists():\n",
    "    ROOT = ROOT.parent\n",
    "sys.path.insert(0, str(ROOT / \"src\"))\n",
    "\n",
    "from ca_personas.compare_agents import (\n",
    "    COMPARISON_METRICS,\n",
    "    run_ml_vs_llm_comparison,\n",
    ")\n",
    "from ca_personas.paths import cohort_source_label, full_cohort_paths\n",
    "from ca_personas.personas import RESEARCH_TIERS\n",
    "\n",
    "PROLIFIC, QUALTRICS = full_cohort_paths()\n",
    "OUT_DIR = ROOT / \"outputs\" / \"ml_vs_llm\"\n",
    "\n",
    "LLM_PROVIDER = os.getenv(\"CA_LLM_PROVIDER\", \"mock\")\n",
    "LLM_MODEL = os.getenv(\"OLLAMA_MODEL\") or os.getenv(\"OPENROUTER_MODEL\") or None\n",
    "TIERS = list(RESEARCH_TIERS)\n",
    "\n",
    "pd.set_option(\"display.max_columns\", 40)\n",
    "print(\"Data source:\", cohort_source_label())\n",
    "print(\"Prolific:\", [str(p) for p in PROLIFIC])\n",
    "print(\"Qualtrics:\", QUALTRICS)\n",
    "print(\"LLM provider:\", LLM_PROVIDER, \"| tiers:\", TIERS)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "04377e20",
   "metadata": {},
   "source": [
    "## 2. Run both agent families\n",
    "\n",
    "1. **ML** \u2014 cross-validated Random Forest + KNN on each tier  \n",
    "2. **LLM** \u2014 persona prompts \u2192 model JSON scores/bands  \n",
    "\n",
    "Both are scored with `evaluate_predictions` / `summarize_errors`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "e0a9c63a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:41.073168Z",
     "iopub.status.busy": "2026-07-25T15:40:41.073000Z",
     "iopub.status.idle": "2026-07-25T15:40:48.904675Z",
     "shell.execute_reply": "2026-07-25T15:40:48.903730Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Artifacts:\n",
      "  predictions  -> /workspace/outputs/ml_vs_llm/agent_predictions.csv\n",
      "  evaluation   -> /workspace/outputs/ml_vs_llm/agent_evaluation.csv\n",
      "  summary      -> /workspace/outputs/ml_vs_llm/agent_summary_by_tier.csv\n",
      "  comparison   -> /workspace/outputs/ml_vs_llm/ml_vs_llm_comparison.csv\n",
      "  deltas       -> /workspace/outputs/ml_vs_llm/ml_vs_llm_deltas.csv\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>agent</th>\n",
       "      <th>agent_family</th>\n",
       "      <th>tier</th>\n",
       "      <th>n_with_ground_truth</th>\n",
       "      <th>mae_group</th>\n",
       "      <th>mae_interpersonal</th>\n",
       "      <th>exact_acc_group</th>\n",
       "      <th>exact_acc_interpersonal</th>\n",
       "      <th>band_acc_group</th>\n",
       "      <th>band_acc_interpersonal</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>demos</td>\n",
       "      <td>241</td>\n",
       "      <td>8.311203</td>\n",
       "      <td>8.298755</td>\n",
       "      <td>0.053942</td>\n",
       "      <td>0.041494</td>\n",
       "      <td>0.311203</td>\n",
       "      <td>0.286307</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>demos</td>\n",
       "      <td>241</td>\n",
       "      <td>5.828067</td>\n",
       "      <td>5.260379</td>\n",
       "      <td>0.033195</td>\n",
       "      <td>0.041494</td>\n",
       "      <td>0.356846</td>\n",
       "      <td>0.394191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>demos</td>\n",
       "      <td>241</td>\n",
       "      <td>5.716706</td>\n",
       "      <td>5.217679</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.091286</td>\n",
       "      <td>0.365145</td>\n",
       "      <td>0.369295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>employment</td>\n",
       "      <td>241</td>\n",
       "      <td>7.912863</td>\n",
       "      <td>7.858921</td>\n",
       "      <td>0.033195</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.323651</td>\n",
       "      <td>0.331950</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>employment</td>\n",
       "      <td>241</td>\n",
       "      <td>5.836886</td>\n",
       "      <td>5.251728</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.369295</td>\n",
       "      <td>0.419087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>employment</td>\n",
       "      <td>241</td>\n",
       "      <td>5.591651</td>\n",
       "      <td>5.106390</td>\n",
       "      <td>0.037344</td>\n",
       "      <td>0.074689</td>\n",
       "      <td>0.365145</td>\n",
       "      <td>0.373444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>geo</td>\n",
       "      <td>241</td>\n",
       "      <td>7.460581</td>\n",
       "      <td>7.908714</td>\n",
       "      <td>0.066390</td>\n",
       "      <td>0.045643</td>\n",
       "      <td>0.381743</td>\n",
       "      <td>0.340249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>geo</td>\n",
       "      <td>241</td>\n",
       "      <td>5.741340</td>\n",
       "      <td>5.004122</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.078838</td>\n",
       "      <td>0.377593</td>\n",
       "      <td>0.452282</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>geo</td>\n",
       "      <td>241</td>\n",
       "      <td>5.045643</td>\n",
       "      <td>4.453983</td>\n",
       "      <td>0.078838</td>\n",
       "      <td>0.107884</td>\n",
       "      <td>0.419087</td>\n",
       "      <td>0.485477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>transit</td>\n",
       "      <td>241</td>\n",
       "      <td>7.419087</td>\n",
       "      <td>8.556017</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.385892</td>\n",
       "      <td>0.302905</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>transit</td>\n",
       "      <td>241</td>\n",
       "      <td>5.209416</td>\n",
       "      <td>4.750352</td>\n",
       "      <td>0.053942</td>\n",
       "      <td>0.087137</td>\n",
       "      <td>0.419087</td>\n",
       "      <td>0.481328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>transit</td>\n",
       "      <td>241</td>\n",
       "      <td>4.676162</td>\n",
       "      <td>4.274793</td>\n",
       "      <td>0.062241</td>\n",
       "      <td>0.074689</td>\n",
       "      <td>0.481328</td>\n",
       "      <td>0.531120</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                agent agent_family        tier  n_with_ground_truth  \\\n",
       "3   mock:mock-persona          llm       demos                  241   \n",
       "4              ml:knn           ml       demos                  241   \n",
       "5    ml:random_forest           ml       demos                  241   \n",
       "6   mock:mock-persona          llm  employment                  241   \n",
       "7              ml:knn           ml  employment                  241   \n",
       "8    ml:random_forest           ml  employment                  241   \n",
       "9   mock:mock-persona          llm         geo                  241   \n",
       "10             ml:knn           ml         geo                  241   \n",
       "11   ml:random_forest           ml         geo                  241   \n",
       "12  mock:mock-persona          llm     transit                  241   \n",
       "13             ml:knn           ml     transit                  241   \n",
       "14   ml:random_forest           ml     transit                  241   \n",
       "\n",
       "    mae_group  mae_interpersonal  exact_acc_group  exact_acc_interpersonal  \\\n",
       "3    8.311203           8.298755         0.053942                 0.041494   \n",
       "4    5.828067           5.260379         0.033195                 0.041494   \n",
       "5    5.716706           5.217679         0.049793                 0.091286   \n",
       "6    7.912863           7.858921         0.033195                 0.049793   \n",
       "7    5.836886           5.251728         0.049793                 0.049793   \n",
       "8    5.591651           5.106390         0.037344                 0.074689   \n",
       "9    7.460581           7.908714         0.066390                 0.045643   \n",
       "10   5.741340           5.004122         0.049793                 0.078838   \n",
       "11   5.045643           4.453983         0.078838                 0.107884   \n",
       "12   7.419087           8.556017         0.049793                 0.049793   \n",
       "13   5.209416           4.750352         0.053942                 0.087137   \n",
       "14   4.676162           4.274793         0.062241                 0.074689   \n",
       "\n",
       "    band_acc_group  band_acc_interpersonal  \n",
       "3         0.311203                0.286307  \n",
       "4         0.356846                0.394191  \n",
       "5         0.365145                0.369295  \n",
       "6         0.323651                0.331950  \n",
       "7         0.369295                0.419087  \n",
       "8         0.365145                0.373444  \n",
       "9         0.381743                0.340249  \n",
       "10        0.377593                0.452282  \n",
       "11        0.419087                0.485477  \n",
       "12        0.385892                0.302905  \n",
       "13        0.419087                0.481328  \n",
       "14        0.481328                0.531120  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = run_ml_vs_llm_comparison(\n",
    "    PROLIFIC,\n",
    "    QUALTRICS,\n",
    "    tiers=TIERS,\n",
    "    llm_provider=LLM_PROVIDER,\n",
    "    llm_model=LLM_MODEL,\n",
    "    join_how=\"inner\",\n",
    "    output_dir=OUT_DIR,\n",
    ")\n",
    "\n",
    "comparison = result[\"comparison\"]\n",
    "deltas = result[\"deltas\"]\n",
    "evaluation = result[\"evaluation\"]\n",
    "\n",
    "print(\"Artifacts:\")\n",
    "for k, p in result[\"artifacts\"].items():\n",
    "    print(f\"  {k:12s} -> {p}\")\n",
    "\n",
    "comparison[comparison[\"tier\"] != \"all\"][\n",
    "    [\"agent\", \"agent_family\", \"tier\", \"n_with_ground_truth\"]\n",
    "    + [c for c in COMPARISON_METRICS if c in comparison.columns][:6]\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b9104445",
   "metadata": {},
   "source": [
    "## 3. Head-to-head metric table\n",
    "\n",
    "Lower is better for MAE / distance metrics; higher is better for exact/band accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f1ad8c66",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:48.906163Z",
     "iopub.status.busy": "2026-07-25T15:40:48.906056Z",
     "iopub.status.idle": "2026-07-25T15:40:48.914918Z",
     "shell.execute_reply": "2026-07-25T15:40:48.913980Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>agent</th>\n",
       "      <th>agent_family</th>\n",
       "      <th>tier</th>\n",
       "      <th>mae_group</th>\n",
       "      <th>mae_interpersonal</th>\n",
       "      <th>exact_acc_group</th>\n",
       "      <th>exact_acc_interpersonal</th>\n",
       "      <th>band_acc_group</th>\n",
       "      <th>band_acc_interpersonal</th>\n",
       "      <th>mean_norm_score_distance_group</th>\n",
       "      <th>mean_band_distance_group</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>demos</td>\n",
       "      <td>8.311203</td>\n",
       "      <td>8.298755</td>\n",
       "      <td>0.053942</td>\n",
       "      <td>0.041494</td>\n",
       "      <td>0.311203</td>\n",
       "      <td>0.286307</td>\n",
       "      <td>0.346300</td>\n",
       "      <td>1.037344</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>demos</td>\n",
       "      <td>5.716706</td>\n",
       "      <td>5.217679</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.091286</td>\n",
       "      <td>0.365145</td>\n",
       "      <td>0.369295</td>\n",
       "      <td>0.238196</td>\n",
       "      <td>0.780083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>demos</td>\n",
       "      <td>5.828067</td>\n",
       "      <td>5.260379</td>\n",
       "      <td>0.033195</td>\n",
       "      <td>0.041494</td>\n",
       "      <td>0.356846</td>\n",
       "      <td>0.394191</td>\n",
       "      <td>0.242836</td>\n",
       "      <td>0.775934</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>employment</td>\n",
       "      <td>7.912863</td>\n",
       "      <td>7.858921</td>\n",
       "      <td>0.033195</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.323651</td>\n",
       "      <td>0.331950</td>\n",
       "      <td>0.329703</td>\n",
       "      <td>0.970954</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>employment</td>\n",
       "      <td>5.591651</td>\n",
       "      <td>5.106390</td>\n",
       "      <td>0.037344</td>\n",
       "      <td>0.074689</td>\n",
       "      <td>0.365145</td>\n",
       "      <td>0.373444</td>\n",
       "      <td>0.232985</td>\n",
       "      <td>0.759336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>employment</td>\n",
       "      <td>5.836886</td>\n",
       "      <td>5.251728</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.369295</td>\n",
       "      <td>0.419087</td>\n",
       "      <td>0.243204</td>\n",
       "      <td>0.784232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>geo</td>\n",
       "      <td>7.460581</td>\n",
       "      <td>7.908714</td>\n",
       "      <td>0.066390</td>\n",
       "      <td>0.045643</td>\n",
       "      <td>0.381743</td>\n",
       "      <td>0.340249</td>\n",
       "      <td>0.310858</td>\n",
       "      <td>0.883817</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>geo</td>\n",
       "      <td>5.045643</td>\n",
       "      <td>4.453983</td>\n",
       "      <td>0.078838</td>\n",
       "      <td>0.107884</td>\n",
       "      <td>0.419087</td>\n",
       "      <td>0.485477</td>\n",
       "      <td>0.210235</td>\n",
       "      <td>0.680498</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>geo</td>\n",
       "      <td>5.741340</td>\n",
       "      <td>5.004122</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.078838</td>\n",
       "      <td>0.377593</td>\n",
       "      <td>0.452282</td>\n",
       "      <td>0.239223</td>\n",
       "      <td>0.780083</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>transit</td>\n",
       "      <td>7.419087</td>\n",
       "      <td>8.556017</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.049793</td>\n",
       "      <td>0.385892</td>\n",
       "      <td>0.302905</td>\n",
       "      <td>0.309129</td>\n",
       "      <td>0.875519</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>transit</td>\n",
       "      <td>4.676162</td>\n",
       "      <td>4.274793</td>\n",
       "      <td>0.062241</td>\n",
       "      <td>0.074689</td>\n",
       "      <td>0.481328</td>\n",
       "      <td>0.531120</td>\n",
       "      <td>0.194840</td>\n",
       "      <td>0.572614</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>transit</td>\n",
       "      <td>5.209416</td>\n",
       "      <td>4.750352</td>\n",
       "      <td>0.053942</td>\n",
       "      <td>0.087137</td>\n",
       "      <td>0.419087</td>\n",
       "      <td>0.481328</td>\n",
       "      <td>0.217059</td>\n",
       "      <td>0.676349</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                agent agent_family        tier  mae_group  mae_interpersonal  \\\n",
       "3   mock:mock-persona          llm       demos   8.311203           8.298755   \n",
       "5    ml:random_forest           ml       demos   5.716706           5.217679   \n",
       "4              ml:knn           ml       demos   5.828067           5.260379   \n",
       "6   mock:mock-persona          llm  employment   7.912863           7.858921   \n",
       "8    ml:random_forest           ml  employment   5.591651           5.106390   \n",
       "7              ml:knn           ml  employment   5.836886           5.251728   \n",
       "9   mock:mock-persona          llm         geo   7.460581           7.908714   \n",
       "11   ml:random_forest           ml         geo   5.045643           4.453983   \n",
       "10             ml:knn           ml         geo   5.741340           5.004122   \n",
       "12  mock:mock-persona          llm     transit   7.419087           8.556017   \n",
       "14   ml:random_forest           ml     transit   4.676162           4.274793   \n",
       "13             ml:knn           ml     transit   5.209416           4.750352   \n",
       "\n",
       "    exact_acc_group  exact_acc_interpersonal  band_acc_group  \\\n",
       "3          0.053942                 0.041494        0.311203   \n",
       "5          0.049793                 0.091286        0.365145   \n",
       "4          0.033195                 0.041494        0.356846   \n",
       "6          0.033195                 0.049793        0.323651   \n",
       "8          0.037344                 0.074689        0.365145   \n",
       "7          0.049793                 0.049793        0.369295   \n",
       "9          0.066390                 0.045643        0.381743   \n",
       "11         0.078838                 0.107884        0.419087   \n",
       "10         0.049793                 0.078838        0.377593   \n",
       "12         0.049793                 0.049793        0.385892   \n",
       "14         0.062241                 0.074689        0.481328   \n",
       "13         0.053942                 0.087137        0.419087   \n",
       "\n",
       "    band_acc_interpersonal  mean_norm_score_distance_group  \\\n",
       "3                 0.286307                        0.346300   \n",
       "5                 0.369295                        0.238196   \n",
       "4                 0.394191                        0.242836   \n",
       "6                 0.331950                        0.329703   \n",
       "8                 0.373444                        0.232985   \n",
       "7                 0.419087                        0.243204   \n",
       "9                 0.340249                        0.310858   \n",
       "11                0.485477                        0.210235   \n",
       "10                0.452282                        0.239223   \n",
       "12                0.302905                        0.309129   \n",
       "14                0.531120                        0.194840   \n",
       "13                0.481328                        0.217059   \n",
       "\n",
       "    mean_band_distance_group  \n",
       "3                   1.037344  \n",
       "5                   0.780083  \n",
       "4                   0.775934  \n",
       "6                   0.970954  \n",
       "8                   0.759336  \n",
       "7                   0.784232  \n",
       "9                   0.883817  \n",
       "11                  0.680498  \n",
       "10                  0.780083  \n",
       "12                  0.875519  \n",
       "14                  0.572614  \n",
       "13                  0.676349  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tier_compare = comparison[comparison[\"tier\"] != \"all\"].copy()\n",
    "display_cols = [\n",
    "    \"agent\",\n",
    "    \"agent_family\",\n",
    "    \"tier\",\n",
    "    \"mae_group\",\n",
    "    \"mae_interpersonal\",\n",
    "    \"exact_acc_group\",\n",
    "    \"exact_acc_interpersonal\",\n",
    "    \"band_acc_group\",\n",
    "    \"band_acc_interpersonal\",\n",
    "    \"mean_norm_score_distance_group\",\n",
    "    \"mean_band_distance_group\",\n",
    "]\n",
    "tier_compare[[c for c in display_cols if c in tier_compare.columns]].sort_values(\n",
    "    [\"tier\", \"agent_family\", \"mae_group\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c0041cd9",
   "metadata": {},
   "source": [
    "## 4. Visual comparison by tier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "cce40f97",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:48.916358Z",
     "iopub.status.busy": "2026-07-25T15:40:48.916227Z",
     "iopub.status.idle": "2026-07-25T15:40:49.170742Z",
     "shell.execute_reply": "2026-07-25T15:40:49.170049Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x800 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "metrics_to_plot = [\n",
    "    (\"mae_group\", \"MAE \u2014 Group CA (\u2193 better)\"),\n",
    "    (\"mae_interpersonal\", \"MAE \u2014 Interpersonal CA (\u2193 better)\"),\n",
    "    (\"band_acc_group\", \"Band accuracy \u2014 Group (\u2191 better)\"),\n",
    "    (\"mean_norm_score_distance_group\", \"Norm. score distance \u2014 Group (\u2193 better)\"),\n",
    "]\n",
    "\n",
    "agents = sorted(tier_compare[\"agent\"].unique())\n",
    "colors = {\n",
    "    a: (\"#C5050C\" if str(a).startswith(\"ml:\") else \"#333333\")\n",
    "    for a in agents\n",
    "}\n",
    "\n",
    "fig, axes = plt.subplots(2, 2, figsize=(12, 8), sharex=True)\n",
    "axes = axes.ravel()\n",
    "for ax, (metric, title) in zip(axes, metrics_to_plot):\n",
    "    for agent in agents:\n",
    "        frame = tier_compare[tier_compare[\"agent\"] == agent].sort_values(\"tier\")\n",
    "        if metric not in frame.columns:\n",
    "            continue\n",
    "        ax.plot(\n",
    "            frame[\"tier\"],\n",
    "            frame[metric],\n",
    "            marker=\"o\",\n",
    "            label=agent,\n",
    "            color=colors.get(agent),\n",
    "        )\n",
    "    ax.set_title(title)\n",
    "    ax.set_xlabel(\"Persona / feature tier\")\n",
    "    ax.tick_params(axis=\"x\", rotation=20)\n",
    "\n",
    "handles, labels = axes[0].get_legend_handles_labels()\n",
    "fig.legend(handles, labels, loc=\"upper center\", ncol=min(4, len(labels)), frameon=False)\n",
    "fig.suptitle(\"ML vs LLM performance across information tiers\", y=1.02)\n",
    "fig.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "85afff26",
   "metadata": {},
   "source": [
    "## 5. Deltas vs best ML baseline\n",
    "\n",
    "For error/distance metrics, **negative** `delta_vs_best_ml` means the agent beats the strongest ML model on that tier. For accuracy metrics, **positive** deltas are better."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "736f8d77",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:49.172351Z",
     "iopub.status.busy": "2026-07-25T15:40:49.172242Z",
     "iopub.status.idle": "2026-07-25T15:40:49.179877Z",
     "shell.execute_reply": "2026-07-25T15:40:49.179295Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tier</th>\n",
       "      <th>agent</th>\n",
       "      <th>agent_family</th>\n",
       "      <th>metric</th>\n",
       "      <th>value</th>\n",
       "      <th>baseline_value</th>\n",
       "      <th>delta_vs_best_ml</th>\n",
       "      <th>better_direction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>demos</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>8.311203</td>\n",
       "      <td>5.716706</td>\n",
       "      <td>2.594498</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>demos</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.828067</td>\n",
       "      <td>5.716706</td>\n",
       "      <td>0.111361</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>demos</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.716706</td>\n",
       "      <td>5.716706</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>employment</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>7.912863</td>\n",
       "      <td>5.591651</td>\n",
       "      <td>2.321212</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>employment</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.836886</td>\n",
       "      <td>5.591651</td>\n",
       "      <td>0.245235</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>employment</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.591651</td>\n",
       "      <td>5.591651</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>geo</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>7.460581</td>\n",
       "      <td>5.045643</td>\n",
       "      <td>2.414938</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>geo</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.741340</td>\n",
       "      <td>5.045643</td>\n",
       "      <td>0.695697</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>geo</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.045643</td>\n",
       "      <td>5.045643</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>transit</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>7.419087</td>\n",
       "      <td>4.676162</td>\n",
       "      <td>2.742925</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>transit</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>5.209416</td>\n",
       "      <td>4.676162</td>\n",
       "      <td>0.533255</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>transit</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_group</td>\n",
       "      <td>4.676162</td>\n",
       "      <td>4.676162</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>demos</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>8.298755</td>\n",
       "      <td>5.217679</td>\n",
       "      <td>3.081076</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>demos</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>5.260379</td>\n",
       "      <td>5.217679</td>\n",
       "      <td>0.042700</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>demos</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>5.217679</td>\n",
       "      <td>5.217679</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>employment</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>7.858921</td>\n",
       "      <td>5.106390</td>\n",
       "      <td>2.752532</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>employment</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>5.251728</td>\n",
       "      <td>5.106390</td>\n",
       "      <td>0.145338</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>employment</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>5.106390</td>\n",
       "      <td>5.106390</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>geo</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>7.908714</td>\n",
       "      <td>4.453983</td>\n",
       "      <td>3.454730</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>geo</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>5.004122</td>\n",
       "      <td>4.453983</td>\n",
       "      <td>0.550139</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>geo</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>4.453983</td>\n",
       "      <td>4.453983</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>transit</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>8.556017</td>\n",
       "      <td>4.274793</td>\n",
       "      <td>4.281224</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>transit</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>4.750352</td>\n",
       "      <td>4.274793</td>\n",
       "      <td>0.475559</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>transit</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>mae_interpersonal</td>\n",
       "      <td>4.274793</td>\n",
       "      <td>4.274793</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>lower</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          tier              agent agent_family             metric     value  \\\n",
       "0        demos  mock:mock-persona          llm          mae_group  8.311203   \n",
       "1        demos             ml:knn           ml          mae_group  5.828067   \n",
       "2        demos   ml:random_forest           ml          mae_group  5.716706   \n",
       "3   employment  mock:mock-persona          llm          mae_group  7.912863   \n",
       "4   employment             ml:knn           ml          mae_group  5.836886   \n",
       "5   employment   ml:random_forest           ml          mae_group  5.591651   \n",
       "6          geo  mock:mock-persona          llm          mae_group  7.460581   \n",
       "7          geo             ml:knn           ml          mae_group  5.741340   \n",
       "8          geo   ml:random_forest           ml          mae_group  5.045643   \n",
       "9      transit  mock:mock-persona          llm          mae_group  7.419087   \n",
       "10     transit             ml:knn           ml          mae_group  5.209416   \n",
       "11     transit   ml:random_forest           ml          mae_group  4.676162   \n",
       "12       demos  mock:mock-persona          llm  mae_interpersonal  8.298755   \n",
       "13       demos             ml:knn           ml  mae_interpersonal  5.260379   \n",
       "14       demos   ml:random_forest           ml  mae_interpersonal  5.217679   \n",
       "15  employment  mock:mock-persona          llm  mae_interpersonal  7.858921   \n",
       "16  employment             ml:knn           ml  mae_interpersonal  5.251728   \n",
       "17  employment   ml:random_forest           ml  mae_interpersonal  5.106390   \n",
       "18         geo  mock:mock-persona          llm  mae_interpersonal  7.908714   \n",
       "19         geo             ml:knn           ml  mae_interpersonal  5.004122   \n",
       "20         geo   ml:random_forest           ml  mae_interpersonal  4.453983   \n",
       "21     transit  mock:mock-persona          llm  mae_interpersonal  8.556017   \n",
       "22     transit             ml:knn           ml  mae_interpersonal  4.750352   \n",
       "23     transit   ml:random_forest           ml  mae_interpersonal  4.274793   \n",
       "\n",
       "    baseline_value  delta_vs_best_ml better_direction  \n",
       "0         5.716706          2.594498            lower  \n",
       "1         5.716706          0.111361            lower  \n",
       "2         5.716706          0.000000            lower  \n",
       "3         5.591651          2.321212            lower  \n",
       "4         5.591651          0.245235            lower  \n",
       "5         5.591651          0.000000            lower  \n",
       "6         5.045643          2.414938            lower  \n",
       "7         5.045643          0.695697            lower  \n",
       "8         5.045643          0.000000            lower  \n",
       "9         4.676162          2.742925            lower  \n",
       "10        4.676162          0.533255            lower  \n",
       "11        4.676162          0.000000            lower  \n",
       "12        5.217679          3.081076            lower  \n",
       "13        5.217679          0.042700            lower  \n",
       "14        5.217679          0.000000            lower  \n",
       "15        5.106390          2.752532            lower  \n",
       "16        5.106390          0.145338            lower  \n",
       "17        5.106390          0.000000            lower  \n",
       "18        4.453983          3.454730            lower  \n",
       "19        4.453983          0.550139            lower  \n",
       "20        4.453983          0.000000            lower  \n",
       "21        4.274793          4.281224            lower  \n",
       "22        4.274793          0.475559            lower  \n",
       "23        4.274793          0.000000            lower  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "delta_view = deltas.copy()\n",
    "delta_view.head(24)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "e3ea07aa",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:49.180971Z",
     "iopub.status.busy": "2026-07-25T15:40:49.180828Z",
     "iopub.status.idle": "2026-07-25T15:40:49.263446Z",
     "shell.execute_reply": "2026-07-25T15:40:49.262721Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 900x450 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mae_delta = deltas[deltas[\"metric\"] == \"mae_group\"]\n",
    "fig, ax = plt.subplots(figsize=(9, 4.5))\n",
    "for family, frame in mae_delta.groupby(\"agent_family\"):\n",
    "    ax.scatter(frame[\"tier\"], frame[\"delta_vs_best_ml\"], label=family, s=70, alpha=0.85)\n",
    "ax.axhline(0, color=\"gray\", linestyle=\"--\", linewidth=1)\n",
    "ax.set_ylabel(\"\u0394 MAE group vs best ML\")\n",
    "ax.set_xlabel(\"Tier\")\n",
    "ax.set_title(\"Group-CA MAE relative to best ML baseline (negative = better than ML)\")\n",
    "ax.legend(frameon=False)\n",
    "fig.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da5af1f9",
   "metadata": {},
   "source": [
    "## 6. Per-participant residuals (optional deep dive)\n",
    "\n",
    "Inspect whether LLM misses concentrate on the same participants as ML."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9396e16b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2026-07-25T15:40:49.264613Z",
     "iopub.status.busy": "2026-07-25T15:40:49.264509Z",
     "iopub.status.idle": "2026-07-25T15:40:49.274323Z",
     "shell.execute_reply": "2026-07-25T15:40:49.273630Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>participant_id</th>\n",
       "      <th>tier</th>\n",
       "      <th>agent</th>\n",
       "      <th>agent_family</th>\n",
       "      <th>pred_group_ca</th>\n",
       "      <th>gt_group_ca</th>\n",
       "      <th>abs_error_group</th>\n",
       "      <th>band_match_group</th>\n",
       "      <th>band_distance_group</th>\n",
       "      <th>norm_score_distance_group</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2392</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>demos</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>26.000000</td>\n",
       "      <td>7</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.791667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>demos</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>9.250000</td>\n",
       "      <td>7</td>\n",
       "      <td>2.250000</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.093750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>demos</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>10.620929</td>\n",
       "      <td>7</td>\n",
       "      <td>3.620929</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.150872</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2393</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>employment</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>7</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.708333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>employment</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>8.545455</td>\n",
       "      <td>7</td>\n",
       "      <td>1.545455</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.064394</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>employment</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>11.329429</td>\n",
       "      <td>7</td>\n",
       "      <td>4.329429</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.180393</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2394</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>geo</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>geo</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>10.449300</td>\n",
       "      <td>7</td>\n",
       "      <td>3.449300</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.143721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>geo</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>11.980000</td>\n",
       "      <td>7</td>\n",
       "      <td>4.980000</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.207500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2395</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>transit</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>7</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>transit</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>9.342824</td>\n",
       "      <td>7</td>\n",
       "      <td>2.342824</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.097618</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...</td>\n",
       "      <td>transit</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>11.135000</td>\n",
       "      <td>7</td>\n",
       "      <td>4.135000</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.172292</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2364</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>demos</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>23.000000</td>\n",
       "      <td>23</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>demos</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>13.500000</td>\n",
       "      <td>23</td>\n",
       "      <td>9.500000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.395833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>demos</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>14.856000</td>\n",
       "      <td>23</td>\n",
       "      <td>8.144000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.339333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2365</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>employment</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>23</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.166667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>employment</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>18.727273</td>\n",
       "      <td>23</td>\n",
       "      <td>4.272727</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.178030</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>employment</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>18.980000</td>\n",
       "      <td>23</td>\n",
       "      <td>4.020000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.167500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2366</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>geo</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>29.000000</td>\n",
       "      <td>23</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>geo</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>15.176542</td>\n",
       "      <td>23</td>\n",
       "      <td>7.823458</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.325977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>geo</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>14.865000</td>\n",
       "      <td>23</td>\n",
       "      <td>8.135000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.338958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2367</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>transit</td>\n",
       "      <td>mock:mock-persona</td>\n",
       "      <td>llm</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>23</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>2</td>\n",
       "      <td>0.416667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>transit</td>\n",
       "      <td>ml:knn</td>\n",
       "      <td>ml</td>\n",
       "      <td>20.276731</td>\n",
       "      <td>23</td>\n",
       "      <td>2.723269</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "      <td>0.113470</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...</td>\n",
       "      <td>transit</td>\n",
       "      <td>ml:random_forest</td>\n",
       "      <td>ml</td>\n",
       "      <td>17.680000</td>\n",
       "      <td>23</td>\n",
       "      <td>5.320000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>0.221667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                         participant_id        tier  \\\n",
       "2392  01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...       demos   \n",
       "0     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...       demos   \n",
       "1     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...       demos   \n",
       "2393  01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...  employment   \n",
       "2     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...  employment   \n",
       "3     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...  employment   \n",
       "2394  01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...         geo   \n",
       "4     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...         geo   \n",
       "5     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...         geo   \n",
       "2395  01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...     transit   \n",
       "6     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...     transit   \n",
       "7     01ed97565030e4d13eba2ecba6f408ba9b1d77f47fa2cc...     transit   \n",
       "2364  0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...       demos   \n",
       "8     0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...       demos   \n",
       "9     0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...       demos   \n",
       "2365  0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...  employment   \n",
       "10    0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...  employment   \n",
       "11    0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...  employment   \n",
       "2366  0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...         geo   \n",
       "12    0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...         geo   \n",
       "13    0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...         geo   \n",
       "2367  0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...     transit   \n",
       "14    0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...     transit   \n",
       "15    0249e61e3a173da6ac2e5417f91dd474d39bec678466a1...     transit   \n",
       "\n",
       "                  agent agent_family  pred_group_ca  gt_group_ca  \\\n",
       "2392  mock:mock-persona          llm      26.000000            7   \n",
       "0                ml:knn           ml       9.250000            7   \n",
       "1      ml:random_forest           ml      10.620929            7   \n",
       "2393  mock:mock-persona          llm      24.000000            7   \n",
       "2                ml:knn           ml       8.545455            7   \n",
       "3      ml:random_forest           ml      11.329429            7   \n",
       "2394  mock:mock-persona          llm       7.000000            7   \n",
       "4                ml:knn           ml      10.449300            7   \n",
       "5      ml:random_forest           ml      11.980000            7   \n",
       "2395  mock:mock-persona          llm      19.000000            7   \n",
       "6                ml:knn           ml       9.342824            7   \n",
       "7      ml:random_forest           ml      11.135000            7   \n",
       "2364  mock:mock-persona          llm      23.000000           23   \n",
       "8                ml:knn           ml      13.500000           23   \n",
       "9      ml:random_forest           ml      14.856000           23   \n",
       "2365  mock:mock-persona          llm      19.000000           23   \n",
       "10               ml:knn           ml      18.727273           23   \n",
       "11     ml:random_forest           ml      18.980000           23   \n",
       "2366  mock:mock-persona          llm      29.000000           23   \n",
       "12               ml:knn           ml      15.176542           23   \n",
       "13     ml:random_forest           ml      14.865000           23   \n",
       "2367  mock:mock-persona          llm      13.000000           23   \n",
       "14               ml:knn           ml      20.276731           23   \n",
       "15     ml:random_forest           ml      17.680000           23   \n",
       "\n",
       "      abs_error_group  band_match_group  band_distance_group  \\\n",
       "2392        19.000000             False                    2   \n",
       "0            2.250000              True                    0   \n",
       "1            3.620929              True                    0   \n",
       "2393        17.000000             False                    2   \n",
       "2            1.545455              True                    0   \n",
       "3            4.329429              True                    0   \n",
       "2394         0.000000              True                    0   \n",
       "4            3.449300              True                    0   \n",
       "5            4.980000              True                    0   \n",
       "2395        12.000000             False                    1   \n",
       "6            2.342824              True                    0   \n",
       "7            4.135000              True                    0   \n",
       "2364         0.000000              True                    0   \n",
       "8            9.500000             False                    1   \n",
       "9            8.144000             False                    1   \n",
       "2365         4.000000             False                    1   \n",
       "10           4.272727             False                    1   \n",
       "11           4.020000             False                    1   \n",
       "2366         6.000000              True                    0   \n",
       "12           7.823458             False                    1   \n",
       "13           8.135000             False                    1   \n",
       "2367        10.000000             False                    2   \n",
       "14           2.723269              True                    0   \n",
       "15           5.320000             False                    1   \n",
       "\n",
       "      norm_score_distance_group  \n",
       "2392                   0.791667  \n",
       "0                      0.093750  \n",
       "1                      0.150872  \n",
       "2393                   0.708333  \n",
       "2                      0.064394  \n",
       "3                      0.180393  \n",
       "2394                   0.000000  \n",
       "4                      0.143721  \n",
       "5                      0.207500  \n",
       "2395                   0.500000  \n",
       "6                      0.097618  \n",
       "7                      0.172292  \n",
       "2364                   0.000000  \n",
       "8                      0.395833  \n",
       "9                      0.339333  \n",
       "2365                   0.166667  \n",
       "10                     0.178030  \n",
       "11                     0.167500  \n",
       "2366                   0.250000  \n",
       "12                     0.325977  \n",
       "13                     0.338958  \n",
       "2367                   0.416667  \n",
       "14                     0.113470  \n",
       "15                     0.221667  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resid_cols = [\n",
    "    c\n",
    "    for c in [\n",
    "        \"participant_id\",\n",
    "        \"tier\",\n",
    "        \"agent\",\n",
    "        \"agent_family\",\n",
    "        \"pred_group_ca\",\n",
    "        \"gt_group_ca\",\n",
    "        \"abs_error_group\",\n",
    "        \"band_match_group\",\n",
    "        \"band_distance_group\",\n",
    "        \"norm_score_distance_group\",\n",
    "    ]\n",
    "    if c in evaluation.columns\n",
    "]\n",
    "evaluation[resid_cols].sort_values([\"participant_id\", \"tier\", \"agent_family\"]).head(24)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de24e968",
   "metadata": {},
   "source": [
    "## 7. Interpretation checklist\n",
    "\n",
    "1. Does any LLM tier beat RF/KNN on **MAE** or **normalized score distance**?\n",
    "2. Does the LLM win on **band accuracy** even when exact scores miss (near-miss viability)?\n",
    "3. As tiers add employment / geo / transit, do ML and LLM improve at the same rate?\n",
    "4. Replace `mock` with Ollama/OpenRouter for the manuscript comparison; the metric definitions stay fixed.\n",
    "\n",
    "```bash\n",
    "CA_LLM_PROVIDER=ollama OLLAMA_MODEL=llama3.2 \\\n",
    "  jupyter nbconvert --to notebook --execute notebooks/ml_vs_llm_comparison.ipynb\n",
    "```"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (.venv psych755-jjb)",
   "language": "python",
   "name": "psych755-jjb"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}