{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "cc02f682",
   "metadata": {},
   "source": [
    "# Multi-Factor / DOE Plots\n",
    "\n",
    "Multi-factor EDA analyses designed experiments (DOE) where multiple\n",
    "factors are varied simultaneously.  Pass named factor arrays via the\n",
    "`factors` dict to `EDAData`.\n",
    "\n",
    "The plots answer: *which factors drive location or spread,\n",
    "and what does the response surface look like?*\n",
    "\n",
    "Reference: [NIST Handbook Chapter 5](https://www.itl.nist.gov/div898/handbook/pri/pri.htm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e4a73452",
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from drippy import EDAData\n",
    "from drippy import (\n",
    "    doe_scatter_plot,\n",
    "    doe_mean_plot,\n",
    "    doe_sd_plot,\n",
    "    contour_plot,\n",
    ")\n",
    "\n",
    "rng = np.random.default_rng(42)\n",
    "\n",
    "# 2-factor full-factorial DOE — 16 runs\n",
    "# Factor A: temperature (+1 = high, −1 = low)\n",
    "# Factor B: pressure (+1 = high, −1 = low)\n",
    "# Response: process yield\n",
    "A = np.tile([-1, 1], 8)\n",
    "B = np.repeat([-1, 1], 8)\n",
    "y = 2.0 * A + 1.5 * B + 0.5 * A * B + rng.normal(scale=0.3, size=16)\n",
    "data = EDAData(y=y, factors={\"Temperature\": A, \"Pressure\": B})"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "50699343",
   "metadata": {},
   "source": [
    "## DOE Scatter Plot\n",
    "\n",
    "One subplot per factor.  Scatters `y` against each factor's coded\n",
    "level to reveal the raw relationship before any modelling."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8b426789",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, axes = doe_scatter_plot(data)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f9dfe065",
   "metadata": {},
   "source": [
    "## DOE Mean Plot\n",
    "\n",
    "For each factor, plots the mean response at each level and draws\n",
    "the grand mean as a reference.  Steep slopes signal large main effects."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b8ac321f",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, axes = doe_mean_plot(data)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6c3ec2e",
   "metadata": {},
   "source": [
    "Temperature has a larger slope than Pressure, indicating a stronger\n",
    "effect on yield.\n",
    "\n",
    "## DOE SD Plot\n",
    "\n",
    "Like the mean plot but for within-level standard deviations.\n",
    "Detects factors that affect process *variability*, not just location."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4ae339a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, axes = doe_sd_plot(data)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f054e93",
   "metadata": {},
   "source": [
    "## Contour Plot  (NIST 1.3.3.10)\n",
    "\n",
    "Interpolates a contour surface over the 2-factor design space.\n",
    "Requires exactly two factors.  Optimal conditions appear as\n",
    "peaks or ridges in the surface."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "91c3831a",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, ax = contour_plot(data)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9dd9996f",
   "metadata": {},
   "source": [
    "Overlay the actual design points:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cdaa0434",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig, ax = contour_plot(data, doe=True)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": "3.12.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}