{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "# Doctor Recommendations Quickstart\n",
        "\n",
        "This notebook shows how to get `cpd doctor` recommendations, fall back to a bundled snapshot when live CLI execution is unavailable, and run a recommended pipeline in Python."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "import json\n",
        "import subprocess\n",
        "from pathlib import Path\n",
        "\n",
        "import numpy as np\n",
        "import cpd\n",
        "\n",
        "try:\n",
        "    import matplotlib.pyplot as plt\n",
        "    HAS_MATPLOTLIB = True\n",
        "except ImportError:\n",
        "    HAS_MATPLOTLIB = False\n",
        "    print(\"matplotlib is optional. Install with: python -m pip install matplotlib\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Build deterministic seasonal + trend + outlier signal and save CSV for CLI interop\n",
        "rng = np.random.default_rng(2028)\n",
        "t = np.arange(360, dtype=np.float64)\n",
        "seasonal = 6.0 * np.sin(2.0 * np.pi * t / 30.0)\n",
        "trend = 0.03 * t\n",
        "level_shift = np.where(t >= 190.0, 7.5, 0.0)\n",
        "signal = 80.0 + seasonal + trend + level_shift + rng.normal(0.0, 1.3, size=t.shape[0])\n",
        "for idx in (55, 140, 265, 315):\n",
        "    signal[idx] += 10.0\n",
        "\n",
        "notebook_data_dir = Path(\"/Users/admin/Documents/Work/claude-doctor-changepoint/cpd/python/examples/notebooks/data\")\n",
        "notebook_data_dir.mkdir(parents=True, exist_ok=True)\n",
        "doctor_input_csv = notebook_data_dir / \"doctor_series.csv\"\n",
        "np.savetxt(doctor_input_csv, signal.reshape(-1, 1), delimiter=\",\")\n",
        "doctor_input_csv\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Attempt live doctor CLI recommendation generation\n",
        "manifest_path = Path(\"/Users/admin/Documents/Work/claude-doctor-changepoint/cpd/Cargo.toml\")\n",
        "live_output_path = notebook_data_dir / \"doctor_recommendations_live.json\"\n",
        "cmd = [\n",
        "    \"cargo\",\n",
        "    \"run\",\n",
        "    \"--manifest-path\",\n",
        "    str(manifest_path),\n",
        "    \"-p\",\n",
        "    \"cpd-cli\",\n",
        "    \"--\",\n",
        "    \"doctor\",\n",
        "    \"--input\",\n",
        "    str(doctor_input_csv),\n",
        "    \"--objective\",\n",
        "    \"balanced\",\n",
        "    \"--min-confidence\",\n",
        "    \"0.2\",\n",
        "    \"--output\",\n",
        "    str(live_output_path),\n",
        "]\n",
        "\n",
        "doctor_payload = None\n",
        "doctor_source = None\n",
        "doctor_error = None\n",
        "\n",
        "try:\n",
        "    completed = subprocess.run(cmd, check=True, capture_output=True, text=True)\n",
        "    doctor_payload = json.loads(live_output_path.read_text(encoding=\"utf-8\"))\n",
        "    doctor_source = f\"live_cli ({live_output_path})\"\n",
        "    print(\"Live doctor command succeeded.\")\n",
        "    if completed.stdout.strip():\n",
        "        print(completed.stdout.strip())\n",
        "except Exception as exc:\n",
        "    doctor_error = str(exc)\n",
        "    print(\"Live doctor command unavailable; will use fallback snapshot.\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Fallback snapshot path when live CLI is unavailable\n",
        "snapshot_path = Path(\"/Users/admin/Documents/Work/claude-doctor-changepoint/cpd/python/examples/notebooks/data/doctor_recommendations_snapshot.json\")\n",
        "if doctor_payload is None:\n",
        "    doctor_payload = json.loads(snapshot_path.read_text(encoding=\"utf-8\"))\n",
        "    doctor_source = f\"snapshot ({snapshot_path})\"\n",
        "\n",
        "print(\"Recommendation source:\", doctor_source)\n",
        "if doctor_error:\n",
        "    print(\"Live CLI error:\", doctor_error)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Inspect top recommendations\n",
        "recommendations = doctor_payload.get(\"recommendations\", [])\n",
        "if not recommendations:\n",
        "    raise RuntimeError(\"Doctor output did not include any recommendations\")\n",
        "\n",
        "for rec in recommendations[:3]:\n",
        "    print(\n",
        "        f\"rank={rec['rank']} confidence={rec['confidence']:.3f} \"\n",
        "        f\"detector={rec['pipeline']['detector']} cost={rec['pipeline']['cost']}\"\n",
        "    )\n",
        "    if rec[\"warnings\"]:\n",
        "        print(\"  warnings:\", \"; \".join(rec[\"warnings\"]))\n",
        "    print(\"  summary:\", rec[\"explanation\"][\"summary\"])\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Execute top recommended pipeline in Python\n",
        "top_pipeline = recommendations[0][\"pipeline\"]\n",
        "doctor_result = cpd.detect_offline(signal, pipeline=top_pipeline, repro_mode=\"balanced\")\n",
        "print(\"top pipeline breakpoints:\", doctor_result.breakpoints)\n",
        "print(\"algorithm:\", doctor_result.diagnostics.algorithm)\n",
        "print(\"cost_model:\", doctor_result.diagnostics.cost_model)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Optional: visualize the signal and selected breakpoints\n",
        "if HAS_MATPLOTLIB:\n",
        "    fig, ax = plt.subplots(figsize=(12, 4))\n",
        "    ax.plot(signal, color=\"black\", linewidth=1.2, alpha=0.85)\n",
        "    for cp in [bp for bp in doctor_result.breakpoints if bp < len(signal)]:\n",
        "        ax.axvline(cp, color=\"tab:red\", linestyle=\"--\", linewidth=1.5, alpha=0.9)\n",
        "    ax.set_title(\"Doctor-selected pipeline result\")\n",
        "    ax.set_xlabel(\"sample\")\n",
        "    ax.set_ylabel(\"value\")\n",
        "    plt.show()\n",
        "else:\n",
        "    print(\"Skipping plot (matplotlib not installed).\")\n"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.9"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}