{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "641931b5",
   "metadata": {},
   "source": [
    "# Aufgabe 1\n",
    "Arbeiten mit NumPy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "064ce97d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "64c14dbd",
   "metadata": {},
   "outputs": [],
   "source": [
    "W = np.array([[1., 2, 3], [3, 4, 5]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bf259722",
   "metadata": {},
   "outputs": [],
   "source": [
    "W.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "311adc5f",
   "metadata": {},
   "outputs": [],
   "source": [
    "W.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d41d05fe",
   "metadata": {},
   "outputs": [],
   "source": [
    "W.shape[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "911571e0",
   "metadata": {},
   "outputs": [],
   "source": [
    "W.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "aa7d64ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "W.T.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6fee5b46",
   "metadata": {},
   "outputs": [],
   "source": [
    "type(W)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b064a293",
   "metadata": {},
   "outputs": [],
   "source": [
    "type(W.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "915e6f9e",
   "metadata": {},
   "outputs": [],
   "source": [
    "r = np.arange(10)\n",
    "print(r)\n",
    "r.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "716e7295",
   "metadata": {},
   "outputs": [],
   "source": [
    "M = np.arange(12).reshape(3,4)\n",
    "print(M)\n",
    "print(M.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "92baea0a",
   "metadata": {},
   "outputs": [],
   "source": [
    "M[2,0]\n",
    "M[1,2]\n",
    "M[1,:]\n",
    "M[1]\n",
    "M[:,1]\n",
    "M[:,[1]]\n",
    "M[:,[3, 0, 1, 1]]\n",
    "M[:,2:4]\n",
    "M[-2,:]\n",
    "M[-2:,:]\n",
    "M[:,2] = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6d78e777",
   "metadata": {},
   "outputs": [],
   "source": [
    "M > 4\n",
    "M[M > 4]\n",
    "M[M > 4] = -17\n",
    "M"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "27fefbbb",
   "metadata": {
    "lines_to_next_cell": 1
   },
   "outputs": [],
   "source": [
    "s = 0\n",
    "for i in range(M.shape[0]):\n",
    "    for j in range(M.shape[1]):\n",
    "        s += M[i,j]\n",
    "        print(s,',',sep='',end='')\n",
    "    print(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a00b5af1",
   "metadata": {
    "lines_to_next_cell": 1
   },
   "outputs": [],
   "source": [
    "def func(x):\n",
    "    print(x * x)\n",
    "    return x"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99670ae0",
   "metadata": {},
   "source": [
    "# Aufgabe 2\n",
    "Laden und Plotten des Iris-Blüten Datensatzes"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "941aabcc",
   "metadata": {},
   "source": [
    "a) Je Iris-Art gibt es 50 Datenpunkte."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e91507b",
   "metadata": {},
   "source": [
    "b) Laden des Datensatzes mit `loadtxt`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "530bf005",
   "metadata": {},
   "outputs": [],
   "source": [
    "iris_data = np.loadtxt(\"../supplement_files/iris.csv\", delimiter=\",\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "25b3a3c1",
   "metadata": {},
   "source": [
    "c + d) Zerlegen der Daten in Matrizen X und T und den Konventionen anpassen.\n",
    "\n",
    "X = iris_data.T[0:4]\n",
    "T = iris_data.T[4:]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "561b5bc1",
   "metadata": {},
   "source": [
    "e) Scatterplot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "da07641f",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# plt.scatter(X[0:1], X[1:2])\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e673709f",
   "metadata": {},
   "source": [
    "f) Scatterplot mit Farben"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a7d4ead0",
   "metadata": {},
   "outputs": [],
   "source": [
    "from matplotlib import colors\n",
    "\n",
    "cmap = colors.ListedColormap([\"red\", \"green\", \"blue\"])\n",
    "\n",
    "plt.scatter(X[0:1], X[1:2], c=T, cmap=cmap)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1a5307aa",
   "metadata": {},
   "source": [
    "# Aufgabe 3\n",
    "Interaktive 3D Scatter Plots in Python"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "094b1e57",
   "metadata": {},
   "source": [
    "a + b) Iris-Blüten Datensatz mit plotly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "da3b79d5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import plotly.express as px\n",
    "\n",
    "df = px.data.iris()\n",
    "fig = px.scatter_3d(df, x='sepal_length', y='sepal_width', z='petal_width', color='species')\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1466c1b4",
   "metadata": {},
   "source": [
    "c) Plot as html speichern."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4c20a687",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = px.data.iris()\n",
    "fig = px.scatter_3d(df, x='sepal_length', y='sepal_width', z='petal_width', color='species')\n",
    "fig.write_html(\"iris.html\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a71792ac",
   "metadata": {},
   "source": [
    "d) Nur Merkmale zwei bis vier."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5659a0a1",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = px.data.iris()\n",
    "fig = px.scatter_3d(df, x=\"sepal_width\", y=\"petal_length\", z=\"petal_width\", color=\"species\")\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5d1f634",
   "metadata": {},
   "source": [
    "# Aufgabe 4\n",
    "Der Thyroid Disease Datensatz -- ein etwas gröÿerer Datensatz"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "295ec0d0",
   "metadata": {},
   "source": [
    "a) Der Datensatz hat 22 Merkmale."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0d36e655",
   "metadata": {},
   "outputs": [],
   "source": [
    "patient_data = np.loadtxt(\"../supplement_files/thyroid-ann-train.data\")\n",
    "print(len(patient_data[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a313d84d",
   "metadata": {},
   "source": [
    "b) 15 Merkmale sind binär und 6 sind kontinuiertlich. Die Patientengruppe ist nominales Merkmal."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6c34fe8",
   "metadata": {},
   "source": [
    "c) Es gibt drei verschiedene Patientengruppen."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "91a74614",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = patient_data.T[0:21]\n",
    "T = patient_data.T[21:]\n",
    "\n",
    "print(np.unique(T).size)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d6abcd4",
   "metadata": {},
   "source": [
    "d) Anzahl der Patienten pro Gruppe."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8f075e3d",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(np.unique(T, return_counts=True)[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e11160cb",
   "metadata": {},
   "source": [
    "e) Plotten verschiedener Merkmalskombinationen, eingefärbt nach Patientengruppe."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a46c1fc9",
   "metadata": {},
   "source": [
    "i) Merkmal 0, 16, 17"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "60a641ca",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = px.scatter_3d(x=X[0], y=X[16], z=X[17], color=T[0].astype(int).astype(str),\n",
    "                    labels={'x': 'Merkmal 0', 'y': 'Merkmal 16', 'z': 'Merkmal 17', 'color': 'Gruppe'})\n",
    "fig.update_traces(marker={'size': 3})\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "65a3fe5e",
   "metadata": {},
   "source": [
    "ii) Merkmal 0, 17, 18"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2c37db7b",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = px.scatter_3d(x=X[0], y=X[17], z=X[18], color=T[0].astype(int).astype(str),\n",
    "                    labels={'x': 'Merkmal 0', 'y': 'Merkmal 17', 'z': 'Merkmal 18', 'color': 'Gruppe'})\n",
    "fig.update_traces(marker={'size': 3})\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1ff9466",
   "metadata": {},
   "source": [
    "iii) Merkmal 18, 19, 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "72bfc023",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = px.scatter_3d(x=X[18], y=X[19], z=X[20], color=T[0].astype(int).astype(str),\n",
    "                    labels={'x': 'Merkmal 18', 'y': 'Merkmal 19', 'z': 'Merkmal 20', 'color': 'Gruppe'})\n",
    "fig.update_traces(marker={'size': 3})\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed9f06a1",
   "metadata": {},
   "source": [
    "iv) Merkmal 16, 18, 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e90baab5",
   "metadata": {},
   "outputs": [],
   "source": [
    "fig = px.scatter_3d(x=X[16], y=X[18], z=X[20], color=T[0].astype(int).astype(str),\n",
    "                    labels={'x': 'Merkmal 16', 'y': 'Merkmal 18', 'z': 'Merkmal 20', 'color': 'Gruppe'})\n",
    "fig.update_traces(marker={'size': 3})\n",
    "fig.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b3af1bc",
   "metadata": {},
   "source": [
    "f) Am besten geeignet ist vermutlich Kombination iv) (16, 18, 20), danach i) (0, 16, 17).\n",
    "Entscheidend ist Merkmal 16: Gruppe 1 hat hier deutlich höhere Werte als Gruppe 2 und diese\n",
    "wiederum höhere als Gruppe 3. Zusätzlich trennen 18 und 20 Gruppe 1 gut von den anderen.\n",
    "\n",
    "Am schlechtesten geeignet ist ii) (0, 17, 18), da Merkmal 16 fehlt und sich Gruppe 2 und 3\n",
    "dort stark überlappen. iii) (18, 19, 20) trennt Gruppe 1 noch ganz gut ab, Gruppe 2 und 3\n",
    "überlappen aber ebenfalls stark."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "40ac7c0e",
   "metadata": {},
   "source": [
    "g) Ja:\n",
    "- Merkmal 0 (Alter) ist in i) und ii) vermutlich überflüssig, die Werte unterscheiden sich\n",
    "  zwischen den Gruppen kaum.\n",
    "- Merkmal 19 in iii) unterscheidet sich zwischen den Gruppen ebenfalls kaum.\n",
    "- In iv) sind Merkmal 18 und 20 stark korreliert (r ≈ 0.77), eines davon bringt vermutlich\n",
    "  kaum zusätzliche Information."
   ]
  }
 ],
 "metadata": {
  "jupytext": {
   "cell_metadata_filter": "-all",
   "main_language": "python",
   "notebook_metadata_filter": "-all"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}