{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "# !pip install folium scikit-learn scipy" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "import pandas as pd\n", "import folium\n", "import matplotlib.pyplot as plt\n", "from sklearn.cluster import KMeans\n", "from scipy.spatial import distance\n", "%matplotlib inline" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# L'épidémie de choléra à Londres en 1854\n", "\n", "Cette étude porte sur la construction d'une **carte épidémiologique** afin de mieux comprendre l'épidémie de choléra dans le quartier de Soho à Londres en 1854. Par l'analyse des données, nous cherchons à trouver le **centre de l'épidémie** et prouver sa proximité avec l'une des pompes d'une quartier." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Chargement et aperçu des données" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "data_death = pd.read_csv(\"deaths.csv\")\n", "data_pumps = pd.read_csv(\"pumps.csv\")\n", "data_death_pumps = pd.read_csv(\"deaths_and_pumps.csv\")" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "Death dataset columns : ['Death', 'X coordinate', 'Y coordinate']\n", "Pumps dataset columns : ['Pump Name', 'X coordinate', 'Y coordinate']\n", "Death/Pumps dataset columns : ['Number of deaths', 'X coordinate', 'Y coordinate']\n", "\n" ] } ], "source": [ "print(\"\"\"\n", "Death dataset columns : {}\n", "Pumps dataset columns : {}\n", "Death/Pumps dataset columns : {}\n", "\"\"\".format(list(data_death.columns), list(data_pumps.columns), list(data_death_pumps.columns)))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "On renomme les colonnes pour éviter les typos à cause des majuscules et des espaces." ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "death_cols = {\n", " list(data_death.columns)[0]: 'd_count',\n", " list(data_death.columns)[1]: 'x', \n", " list(data_death.columns)[2]: 'y'}\n", "pump_cols = {\n", " list(data_pumps.columns)[0]: 'name',\n", " list(data_pumps.columns)[1]: 'x', \n", " list(data_pumps.columns)[2]: 'y'}\n", "d_p_cols = {\n", " list(data_death_pumps.columns)[0]: 'death_per_pumps',\n", " list(data_death_pumps.columns)[1]: 'x', \n", " list(data_death_pumps.columns)[2]: 'y'}\n", "\n", "data_death.rename(columns=death_cols, inplace=True)\n", "data_pumps.rename(columns=pump_cols, inplace=True)\n", "data_death_pumps.rename(columns=d_p_cols, inplace=True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Un petit regard sur la donnée." ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ " d_count x y\n", "0 1 51.513418 -0.137930\n", "1 1 51.513418 -0.137930\n", "2 1 51.513418 -0.137930\n", "3 1 51.513361 -0.137883\n", "4 1 51.513361 -0.137883\n", "\n", "\n", " name x y\n", "0 Broad St. 51.513341 -0.136668\n", "1 Crown Chapel 51.513876 -0.139586\n", "2 Gt Marlborough 51.514906 -0.139671\n", "3 Dean St. 51.512354 -0.131630\n", "4 So Soho 51.512139 -0.133594\n", "\n", "\n", " death_per_pumps x y\n", "0 3 51.513418 -0.137930\n", "1 2 51.513361 -0.137883\n", "2 1 51.513317 -0.137853\n", "3 1 51.513262 -0.137812\n", "4 4 51.513204 -0.137767\n" ] } ], "source": [ "print(data_death.head())\n", "print('\\n')\n", "print(data_pumps.head())\n", "print('\\n')\n", "print(data_death_pumps.head())" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Donnée manquante dans le dataset death.csv : 0\n", "Donnée manquante dans le dataset pumps.csv : 0\n", "Donnée manquante dans le dataset death_and_pumps.csv : 0\n" ] } ], "source": [ "print(\"Donnée manquante dans le dataset death.csv : {}\".format(len(data_death[data_death.isnull().any(axis=1)])))\n", "print(\"Donnée manquante dans le dataset pumps.csv : {}\".format(len(data_pumps[data_pumps.isnull().any(axis=1)])))\n", "print(\"Donnée manquante dans le dataset death_and_pumps.csv : {}\".format(len(data_death_pumps[data_death_pumps.isnull().any(axis=1)])))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Création de la carte\n", "\n", "### Les décès\n", "\n", "On commence par afficher les décès sur la carte en pointant vers une coordonnée disponible dans le dataset." ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [], "source": [ "data_death_df = data_death.groupby(['x', 'y']).d_count.count().to_frame()\n", "data_death_df.reset_index(inplace=True)\n", "death_coordinates = data_death_df[[\"x\",\"y\"]]\n", "death_coordinates = death_coordinates.values.tolist()" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [], "source": [ "soho_c = death_coordinates[0]\n", "death_map = folium.Map(location=soho_c, tiles='Stamen Toner', zoom_start=17)\n", "for p in range(0, len(death_coordinates)):\n", " folium.CircleMarker(death_coordinates[p], radius=2*int(data_death_df['d_count'][p]), \n", " color='blue', fill=True, fill_color='blue',\n", " opacity = 0.4).add_to(death_map)" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
Make this Notebook Trusted to load map: File -> Trust Notebook
" ], "text/plain": [ "" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" } ], "source": [ "death_map" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Les pompes\n", "\n", "On y ajoute ensuite les emplacements de pompes." ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [], "source": [ "pump_coordinates = data_pumps[[\"x\",\"y\"]]\n", "pump_coordinates = pump_coordinates.values.tolist()" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
Make this Notebook Trusted to load map: File -> Trust Notebook
" ], "text/plain": [ "" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "death_pump_map = death_map\n", "for p in range(0, len(pump_coordinates)):\n", " folium.Marker(pump_coordinates[p],\n", " popup='Name : {}'.format(data_pumps['name'][p]),\n", " icon=folium.Icon(color='red', icon='info-sign')).add_to(death_pump_map)\n", "death_pump_map" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Recherche de la pompe au centre de l'épidémie" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Sur la carte précédente on voit très clairement par un cercle de diamètre supérieur aux autres, que la plus grande densité de décès se trouve au plus près de la pompe de Broad St. Essayons de le démontrer par l'analyse.
\n", "On peut par exemple utiliser l'algorithme [K-means](https://fr.wikipedia.org/wiki/K-moyennes) pour former des **clusters** et vérifier quelle pompe se trouve au centre du cluster contenant le plus de cas.
\n", "On commence par initialiser K-means avec un nombre de clusters correspondant au nombre de pompes." ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [], "source": [ "n_pumps = len(data_pumps)" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [], "source": [ "kmeans = KMeans(n_clusters = n_pumps, init ='k-means++')\n", "kmeans.fit(data_death[data_death.columns[1:3]])\n", "data_death['cluster_label'] = kmeans.fit_predict(data_death[data_death.columns[1:3]])\n", "centers = kmeans.cluster_centers_\n", "labels = kmeans.predict(data_death[data_death.columns[1:3]])" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "On observe la répartition des clusters." ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "data_death.plot.scatter(x = 'x', y = 'y', c=labels, s=20, \n", " ylim=[data_pumps['y'].min()-0.001, data_pumps['y'].max()-0.001],\n", " xlim=[data_pumps['x'].min()+0.0015, data_pumps['x'].max()+0.001], cmap='viridis')\n", "plt.scatter(centers[:, 0], centers[:, 1], c='red', s=100, alpha=0.8)\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Ici on voit que le nombre de clusters est trop grand. Pour certains points, l'appartenance à un cluster plus qu'un autre n'apparaît pas clair. On voit d'ailleurs sur la carte que beaucoup de pompe sont à l'extérieur du centre de l'épidémie.\n", "Il nous faut trouver le nombre optimal de cluster possible. Pour ça on utilise la méthode [Elbow Curve](https://en.wikipedia.org/wiki/Elbow_method_(clustering))" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAacAAAEWCAYAAADCeVhIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3XmcXGWd7/HPt7d0p0nS2XdICAEMO2nD5sKwKDAqKIIwLqAowojo6MxcGOfOeOeOc9EZd3FBcYwbocMyBBGRRUFZJJ0QAgFiQgC7s+9bJ73+7h91Eiqhekmnq0919/f9etWrTp1znlO/05D+9nnqqecoIjAzMyskRWkXYGZmtj+Hk5mZFRyHk5mZFRyHk5mZFRyHk5mZFRyHk5mZFRyHk1keSbpS0h+zXoekI9KsyawvcDiZHSRJr0raJWlH1uM7ade1h6Txkm6VtFrSdkkvSfo/kirTrs2sPQ4ns57x7og4JOtxXdoFAUgaATwJVACnRcQQ4FygCpjWjeOV9GyFZrk5nMx63wWSVkjaIOk/JRUBSCqS9M+SXpO0TtJPJQ1Lts2W9PlkeWLSPfi3yesjJG2SpBzv9TlgO/ChiHgVICLqIuIzEbFY0pTkWHtDR9LvJX08Wb5S0uOSvi5pE/B/JW2RdGzW/qOTK8cxyet3SVqU7PeEpOPz8DO0fs7hZNb73gtUAycDFwIfS9ZfmTz+CjgcOATY0z34KHBmsvx2YEXyDPA24A+Rey6yc4C7IqLtIOo9JXm/McC/AXcBl2dtvxR4NCLWSToZ+DHwSWAk8ANgnqRBB/H+NgA5nMx6xv8kVwp7Hp/oYN8vR8SmiPgL8A1e/0X/QeBrEbEiInYANwKXJVc1jwJvTa6y3gZ8BTgjaff2ZHsuI4HVB3dqrIqIb0dES0TsAn7JvuH0N8k6gE8AP4iIP0VEa0TMBhqBUw+yBhtgHE5mPeOiiKjKevywg33rspZfAyYkyxOS19nbSoCxEfEysAM4EXgr8CtglaSj6DicNgLjD/hs2q8X4BGgQtIpkg5Laro72XYY8PnsoAYm8/o5mnWJw8ms903OWj4UWJUsryLzyz17WwuwNnn9KPB+oCwiViavPwIMBxa1814PAe/d87lWDjuT58FZ68btt88+3YVJF2ENmaunvwF+FRHbk811wJf2C+rBEXFbO+9vlpPDyaz3/YOk4ZImA58Bbk/W3wb8naSpkg4B/gO4PSJaku2PAtcBjyWvfw98GvhjRLS2815fA4YCs5OrnD0DKr4m6fiIWA+sBD4kqVjSx+jaKL5fAh8g0xX5y6z1PwSuSa6qJKlS0l9LGtKFY5rt5XAy6xn37vc9p7s72PceYAGZq537gFuT9T8GfkYmfF4BdpMJnz0eBYbwejj9kcwVz2O0IyI2AacDzcCfJG0HHga2AsuT3T4B/AOZLsBjgCc6O9mI+BOZq64JwP1Z62uT430H2Jy8x5WdHc9sf/LNBs3MrND4ysnMzAqOw8nMzAqOw8nMzAqOw8nMzApOqpM4SjoP+CZQDPwoIm7ab7uS7RcADcCVEbGwo7bJRJe3A1OAV4FLI2Jzsu1G4CqgFbg+Ih5I1s8EfkJmcsxfA59pZyqYvUaNGhVTpkw5qPM3MxtoFixYsCEiRne2X2rhJKkYuJnMDMn1wHxJ8yLihazdzgemJ49TgO8Bp3TS9gbg4Yi4SdINyev/JWkGcBmZobITgIckHZl8P+R7wNXAU2TC6TyyhsfmMmXKFGpra3viR2FmNmBIeq3zvdLt1psFLE/mEWsC5pCZBDPbhcBPI+MpoErS+E7aXgjMTpZnAxdlrZ8TEY0R8QqZ71/MSo43NCKeTK6WfprVxszMUpBmOE1k3zm76pN1Xdmno7ZjI2I1QPI8pgvHqu+kDgAkXS2pVlLt+vXrOzw5MzPrvjTDKde9Z/b/nKe9fbrStqvv1+VjRcQtEVEdEdWjR3faZWpmZt2UZjjVs+8EmJN4fQLMzvbpqO3apKuO5HldF441qZM6zMysF6UZTvOB6ckkl2VkBivM22+fecBHkgkkTwW2Jl11HbWdB1yRLF9BZh6zPesvkzRI0lQygyyeTo63XdKpyejAj2S1MTOzFKQ2Wi8iWiRdBzxAZjj4jyNiiaRrku3fJzNy7gIygxcagI921DY59E1AjaSrgL8AlyRtlkiqAV4gcxuCT2XN5Hwtrw8lv59ORuqZmVl+eeLXbqqurg4PJTczOzCSFkREdWf7pfolXDMzS1dLaxu7mlvZ1dSaeW5upaGpld1Nmec96/Zsb2hq5eyjx3DC5Kq81uVwMjMrUG1twe6WrODICohdzfsGyO6s9bua9t03s62FXc1try8n25tbD7z3bMyQQQ4nM7O+ZFdTK5sbmti0s4ktDc1samhiS0MT23e3tBsgrwdJC7ub2/ZZPlBlxUWUlxYxuKyEirJiKkqLqSgrZnBZCSMPybweXFZM+Z71yfPefUtfX84co4jyPculxZSXFpEZO5ZfDiczsxwigoam1jeEzKadTWxuaGbzziY2NySPnc17lzsKlCLB4LKSJBiKGFxaQnkSEKOHDKKidHBWMGQFSLI8eP8AKds3TCpKiykp7h/zeTuczKzfiwi2N7YkgfJ6sOQKni0NzXufm1pzB40EwypKGTG4jKrBpUyoKmfGhKGMqMy8HjG4jOGVZQwfXMaIylKqBpcxpLyEsuLeueroDxxOZtantLUF23Y3s3lviOQOmddDqJktDU20tOX+bKVIMDwJmRGVZUweMZgTJlVRVZmEzN6gKd0bOMMqSikucsjkk8PJzArC9t3NrN66m5VbdrF6y25Wb93F2m279+tCywRNOzlDSZGoyrpamTb6EIZXlmYCJkfIjEiuaIocNAXH4WRmedfY0sqarbtZlYTOqi27WLV1N6uSIFq1dRfbd7fs06ZIMOqQQYxIguSocUM6DJmqylKGDCpxt1k/4XAys4PS1has39G4zxXPyqzQWbVlNxt2NL6h3fDBpUyoqmDyiMGcevgIxldVMH5YOROrKhhfVcHYIYP6zYf7duAcTmbWrohg665mVm1JrnK2vvGKZ83W3W/4PGdwWTETkrB507ihmeWqciYMq2BCVTnjh1VQUVac0llZX+BwMhvAdjW1smpr9lXOfstbd9PQ1LpPm5IiMW5YOROqKqg+bDjjqyqYUFXBhGGZ0JlYVcHQCnev2cFxOJn1Uy2tbazd3pj5fCcJmszy61dBmxua39Bu9JBBTBhWzpFjh/D2I8cwoap871XQhKoKRh0yyCPVLO8cTmb9QENTC/Nf3cwTyzew4LXNrNySGem2/6i2IeUlmc90hpVz0qFV+4TOhGEVjB02iEEl7m6z9DmczPqgxpZWFv1lC0+8vJEnXt7AorotNLcGpcXihElVnD5tFBOryt8wyOCQQf4nb32D/0816wNa24Ilq7by+PJMGM1/dRO7m9soEhw7cRhXveVwTp82kjdPGeGBBtYvOJzMClBEsGzdDp5YvoHHX97IUys27v0e0JFjD+GyNx/K6dNGcsrUkQwbXJpytWY9z+FkViD+srGBJ17ekHTVbdz73aBDRwzmr48bz2nTRnLatJGMGVKecqVm+edwMkvJum27935m9MTLG6nfvAvIjJZ7yxEjOX3aKE6bNpLJIwanXKlZ73M4mfWSLQ1NPLVi094wWr5uBwBDy0s4bdpIPvHWwznjiJFMG32IvyNkA57DySxPGppaePqVTTz58kYef3kDS1ZtIwIqSouZNXUEl8ycxOnTRjFjwlB/b8hsPw4nsx6yZ3j34y9v5Mn9hnefdOhwPnP2dM44YhQnTKqirMRzxpl1xOFk1k2tbcHzK7fu/dwoe3j3cR7ebXZQHE5mXbRnePfjyzOfGbU7vPvwkQyr8PBus4PhcDJrR0RQt2kXT7y8YW9X3YYdTYCHd5vlm8PJLEv28O7Hl29k5Zbs4d2jPLzbrJc4nGzA293cyjceWsZDL659w/DuT74987mRh3eb9S6Hkw1odZsauObnC1iyahtvnT7Kw7vNCoTDyQasPyxbz6dve4bWtuDWK6o5+01j0y7JzBIOJxtwIoLvP7qC/3zgJY4Ycwg/+HA1U0dVpl2WmWVxONmAsqOxhX+Y+yz3P7+Gvz5+PF+5+HgqfY8js4Ljf5U2YLy8fgef/NkCVqzfwT9dcDSfeOvhHuRgVqAcTjYg/HbJGj5f8yylJUX8/KpTOP2IUWmXZGYdcDhZv9baFnzjoT/z7UeWc9zEYXz/wzOZWFWRdllm1olUZp+UNELSg5KWJc/D29nvPElLJS2XdENX2ku6Mdl/qaR3Zq2fKem5ZNu3lPTnSLpS0npJi5LHx/N57tZ7tjY087GfzOfbjyznkpmTmHvNaQ4msz4iramRbwAejojpwMPJ631IKgZuBs4HZgCXS5rRUftk+2XAMcB5wHeT4wB8D7gamJ48zst6u9sj4sTk8aMePVNLxYurt/Hu7/yRJ17ewL9fdCxfef/xlJd68lWzviKtcLoQmJ0szwYuyrHPLGB5RKyIiCZgTtKuo/YXAnMiojEiXgGWA7MkjQeGRsSTERHAT9t5T+sH7lm0kvd+93EaW1qZc/VpfOjUwzzwwayPSSucxkbEaoDkeUyOfSYCdVmv65N1HbVvr83EZDnXsQAulrRY0h2SJrdXtKSrJdVKql2/fn1n52i9rLm1jX+79wU+M2cRx00cxr2ffgszD8vZY2xmBS5vAyIkPQSMy7HpC109RI510c02HR3rXuC2iGiUdA2ZK7Gzch08Im4BbgGorq7urBbrReu3N3LdLxfyp1c2ceXpU/inC97kG/qZ9WF5C6eIOKe9bZLWShofEauTLrd1OXarB7KvYiYBq5Ll9tq316Y+WX7DsSJiY9b6HwJf7vTkrKAsqtvCNT9bwOaGJr526Qm87+RJnTcys4KW1p+W84ArkuUrgHty7DMfmC5pqqQyMgMd5nXSfh5wmaRBkqaSGfjwdNL1t13SqckovY/saZOE2x7vAV7siRO03jHn6b9w6fefpKRY3Hnt6Q4ms34ire853QTUSLoK+AtwCYCkCcCPIuKCiGiRdB3wAFAM/DgilnTUPiKWSKoBXgBagE9FRGvS5lrgJ0AFcH/yALhe0nuS/TcBV+btrK3HNLa08sV5S7jt6TreOn0U37rsJIZXlqVdlpn1EGUGr9mBqq6ujtra2rTLGJBWb93FNT9fyLN1W/jbM6fx+Xcc5dtbmPURkhZERHVn+3mGCOtTnlqxket+uZBdTa18/0Mnc96x4ztvZGZ9jsPJ+oSI4MePv8p//PpFDhs5mDlXn8oRY4akXZaZ5YnDyQrerqZWbrhrMfcsWsW5M8bytUtPYEh5adplmVkeOZysoP1lYwNX/6yWpWu38/fvOJK/PfMIivz5klm/53CygvW7pev4zG3PIIn/vvLNnHlUrolEzKw/cjhZwWlrC27+3XK+9tCfOXrcUH7woZkcOnJw2mWZWS9yOFlB2ba7mc/XPMuDL6zlwhMncNP7jqeizLOJmw00DicrGMvXbefqny7gtU0N/Mu7ZvDRM6Z4NnGzAcrhZAXhN8+v5vM1z1JRVswvPn4Kpx4+Mu2SzCxFDidLVWtb8F+/Xcr3fv8yJ06u4nsfOpnxw3y3WrOBzuFkqdm8s4nr5zzDH5Zt4PJZh/LF98xgUIk/XzIzh5Ol5PmVW7nm5wtYt62Rm953HJfNOjTtksysgDicrNfdtbCeG+96jhGVZdRccxonTq5KuyQzKzAOJ+s1za1t/PuvXmD2k69xytQR3PzBkxl1yKC0yzKzAuRwsl6xbvtuPvWLhcx/dTMff8tUbjj/aEqKfRt1M8vN4WR5t+C1zVz78wVs293MNy87kQtPnJh2SWZW4BxOljcRwc//9Bf+7d4ljB9WweyPzeJN44emXZaZ9QEOJ8uL3c2t/O//eZ65C+o586jRfPMDJzFssG9zYWZd43CyHrdyyy6u+dkCnlu5levPOoLPnnOkb3NhZgfE4WQ96onlG7jutmdobmnjhx+p5twZY9Muycz6IIeT9YiI4Id/WMFN97/EtNGH8P0Pz2Ta6EPSLsvM+iiHkx20nY0t/OOdi7lv8WouOG4cX3n/CRwyyP9rmVn3+TeIHZRXNuzkmp8tYNm67dxw/tF88m2H+zYXZnbQHE7WbX9ctoFrf7GAkiLx04+dwlumj0q7JDPrJxxO1i1tbcENdy1mzJBB/OSjs5g8wrdRN7Oe4/ljrFueWrGR+s27uP7s6Q4mM+txDifrlpraOoaWl/DOY8alXYqZ9UMOJztgW3c1c//za7jwxImUl/rmgGbW8xxOdsDufXYVjS1tXFo9Oe1SzKyfcjjZAZtbW8fR44Zw7ERP4mpm+eFwsgPy0pptPFu/lUuqJ/v7TGaWNw4nOyBza+spLRYXnTgh7VLMrB9zOFmXNbW08T/PrOScN41lpG+vbmZ5lEo4SRoh6UFJy5Ln4e3sd56kpZKWS7qhK+0l3Zjsv1TSO7PWf0lSnaQd+73HIEm3J23+JGlKz59x//DIS+vYuLPJAyHMLO/SunK6AXg4IqYDDyev9yGpGLgZOB+YAVwuaUZH7ZPtlwHHAOcB302OA3AvMCtHLVcBmyPiCODrwJd75Az7obm1dYwdOoi3epoiM8uztMLpQmB2sjwbuCjHPrOA5RGxIiKagDlJu47aXwjMiYjGiHgFWJ4ch4h4KiJWd1LLHcDZ8if9b7B2225+t3QdF588iZJi9wabWX6l9Vtm7J6gSJ7H5NhnIlCX9bo+WddR+47atGdvm4hoAbYCI3PtKOlqSbWSatevX9/JYfuXuxaupC3gEnfpmVkvyNvEr5IeAnLNbfOFrh4ix7pIs01E3ALcAlBdXd3ZcfuNiGBubR2zpoxg6qjKtMsxswEgb+EUEee0t03SWknjI2K1pPHAuhy71QPZf6ZPAlYly+2176hNe/a0qZdUAgwDNnXSZkBZ8NpmVmzYybVnTku7FDMbINLq1psHXJEsXwHck2Of+cB0SVMllZEZ6DCvk/bzgMuSEXhTgenA0wdQy/uBRyJiwFwVdUVNbR2Dy4q54LjxaZdiZgNEWuF0E3CupGXAuclrJE2Q9GvY+/nPdcADwItATUQs6ah9sr0GeAH4DfCpiGhNjv0VSfXAYEn1kr6YHOtWYKSk5cDnyDFycCDb2djCrxav5l3Hj6fSt143s14iXyR0T3V1ddTW1qZdRt7V1Nbxj3cs5o5rTqN6yoi0yzGzPk7Sgoio7mw/jwm2Dt1RW8/hoyqZeVjO70mbmeWFw8natWL9Dp5+dZMneTWzXudwsnbdsaCe4iJx8cmdfVXMzKxnOZwsp5bWNu5cWM+ZR45mzNDytMsxswHG4WQ5/WHZBtZua/SMEGaWii6Hk6S3SPposjw6+R6R9VM1tXWMrCzjrKNzzSxlZpZfXQonSf8K/C/gxmRVKfDzfBVl6dq4o5GHXlzLRSdNpKzEF9dm1vu6+pvnvcB7gJ0AEbEKGJKvoixd/7NoFc2t4fs2mVlquhpOTcmUPgEgybN/9lN7Jnk9YdIwjhrnvz/MLB1dDacaST8AqiR9AngI+GH+yrK0PL9yGy+t2e6BEGaWqi5NlhYR/yXpXGAbcBTwLxHxYF4rs1TU1NYxqKSId58wIe1SzGwA6zScktucP5DcAsOB1I/tbm7lnkUrOf/YcQyrKE27HDMbwDrt1ktm9W6QNKwX6rEUPbBkDdt2t3gghJmlrqv3QNgNPCfpQZIRewARcX1eqrJUzK2tZ9LwCk49POdd6s3Mek1Xw+m+5GH9VN2mBh5/eQOfPftIioo8yauZpaurAyJmJ3ejPTJZtTQimvNXlvW2OxfWA3DxTE/yambp61I4SToTmA28CgiYLOmKiHgsf6VZb2lrC+bW1nPGtFFMGj447XLMzLrcrfdV4B0RsRRA0pHAbcDMfBVmvefJFRtZuWUX/3jeUWmXYmYGdP1LuKV7ggkgIv5MZn496wdqausYWl7CO48Zl3YpZmZA16+caiXdCvwsef1BYEF+SrLetHVXM795fg2XVk+mvLQ47XLMzICuh9O1wKeA68l85vQY8N18FWW9595nV9HY0ubvNplZQelqOJUA34yIr8HeWSMG5a0q6zVza+s4etwQjp04NO1SzMz26upnTg8DFVmvK8hM/mp92EtrtvFs/VYurZ6M5O82mVnh6Go4lUfEjj0vkmWPOe7j5tbWU1osLjrJ320ys8LS1XDaKenkPS8kVQO78lOS9YamljbufmYl57xpLCMqy9Iux8xsH139zOmzwFxJq8jccHAC8IG8VWV598hLa9m0s8kDIcysIHV45STpzZLGRcR84GjgdqAF+A3wSi/UZ3lSU1vP2KGDeOv0UWmXYmb2Bp116/0AaEqWTwP+CbgZ2Azckse6LI/WbtvN75eu4+KTJ1FS3NWeXTOz3tNZt15xRGxKlj8A3BIRdwJ3SlqU39IsX+5auJK2wLdiN7OC1dmfzcWS9gTY2cAjWdu6+nmVFZCIYG5tHbOmjGDqqMq0yzEzy6mzgLkNeFTSBjKj8/4AIOkIYGuea7M8WPDaZlZs2Mm1Z05LuxQzs3Z1GE4R8SVJDwPjgd9GRCSbioBP57s463k1tXVUlhVzwXHj0y7FzKxdnXbNRcRTOdb9OT/lWD7tbGzhV4tX8+7jJ1A5yL2yZla4UhmqJWmEpAclLUueh7ez33mSlkpaLumGrrSXdGOy/1JJ78xa/yVJdZJ27PceV0paL2lR8vh4Ps65ENz33Goamlq5pHpS2qWYmXUorXHENwAPR8R0MvP23bD/DsnksjcD5wMzgMslzeiofbL9MuAY4Dzgu8lxAO4FZrVTz+0RcWLy+FFPnGAhmltbx+GjKpl5WM6/BczMCkZa4XQhmdu+kzxflGOfWcDyiFgREU3AnKRdR+0vBOZERGNEvAIsT45DRDwVEat7/Ez6iBXrdzD/1c1c4klezawPSCucxu4JiuR5TI59JgJ1Wa/rk3Udte+oTUculrRY0h2S+uWXf+YuqKe4SFx8sid5NbPCl7dPxSU9BOS67/cXunqIHOsix7qDbXMvcFtENEq6hsyV2Fk5Dy5dDVwNcOihh3Zy2MLR0trGnQvqOfPI0YwZWp52OWZmncpbOEXEOe1tk7RW0viIWC1pPLAux271QPZVzCRgVbLcXvuO2rRX58aslz8EvtzBvreQTNtUXV3dWegVjD8s28C67Y2eEcLM+oy0uvXmAVcky1cA9+TYZz4wXdJUSWVkBjrM66T9POAySYMkTQWmA093VEgSbnu8B3jxAM+l4NXU1jGysoyzjs7Ve2pmVnjSCqebgHMlLQPOTV4jaYKkXwNERAtwHfAAmcCoiYglHbVPttcAL5CZOf1TEdGaHPsrkuqBwZLqJX0xOdb1kpZIeha4Hrgyr2feyzbuaOShF9fy3pMmUlbiSV7NrG/Q65M+2IGorq6O2tratMvo1K1/fIX/+6sXeOCzb+OocUPSLsfMBjhJCyKiurP9/Kd0P7ZnktcTJlc5mMysT3E49WPPrdzKS2u2c8lMzwhhZn2Lw6kfq6mtY1BJEe8+YULapZiZHRCHUz+1u7mVexat4vxjxzGsojTtcszMDojDqZ96YMkatu9u4VJ/t8nM+iCHUz9VU1vHpOEVnHr4yLRLMTM7YA6nfqhuUwNPvLyRS2ZOpqjIk7yaWd/jcOqH7lxYD8DFMz3Jq5n1TQ6nfqatLZhbW89bjhjFpOGD0y7HzKxbHE79zJMrNrJyyy5P8mpmfZrDqZ+pqa1jaHkJ75gxNu1SzMy6zeHUj2xtaOb+59dw4YkTKS8t7ryBmVmBcjj1I/MWr6Kppc3fbTKzPs/h1I/Mra3j6HFDOHbi0LRLMTM7KA6nfuKlNdtYXL+VS6snI/m7TWbWtzmc+om5tfWUFouLTvJ3m8ys73M49QNNLW3c/cxKzp0xlhGVZWmXY2Z20BxO/cAjL61l084mf7fJzPoNh1M/UFNbz7ih5bxt+ui0SzEz6xEOpz5u7bbd/H7pOt538kSKPcmrmfUTDqc+7s6F9bQF7tIzs37F4dSHRWQmeZ01ZQRTR1WmXY6ZWY9xOPVhta9t5pUNO7mkelLapZiZ9SiHUx9WM7+OyrJiLjhufNqlmJn1KIdTH7WzsYX7nlvNu46fQOWgkrTLMTPrUQ6nPuq+51bT0NTKpW92l56Z9T8Opz5qbm0dh4+u5ORDh6ddiplZj3M49UEr1u9g/qubPcmrmfVbDqc+aO6CeoqLxPs8yauZ9VMOpz6mpbWNOxfUc+aRoxkztDztcszM8sLh1Mc8tmw967Y3ekYIM+vXHE59TM38ekZWlnHW0WPSLsXMLG8cTn3Ixh2NPPTiWt570kTKSvyfzsz6L/+G60PufmYlLW3hLj0z6/dSCSdJIyQ9KGlZ8pzzyzqSzpO0VNJySTd0pb2kG5P9l0p6Z7JusKT7JL0kaYmkm7L2HyTp9qTNnyRNyd+Zd9+eSV5PmFzFUeOGpF2OmVlepXXldAPwcERMBx5OXu9DUjFwM3A+MAO4XNKMjton2y8DjgHOA76bHAfgvyLiaOAk4AxJ5yfrrwI2R8QRwNeBL/f0yfaE51ZuZena7VzqSV7NbABIK5wuBGYny7OBi3LsMwtYHhErIqIJmJO066j9hcCciGiMiFeA5cCsiGiIiN8BJMdaCEzKcaw7gLNVgN9sramtY1BJEe8+YULapZiZ5V1a4TQ2IlYDJM+5hp5NBOqyXtcn6zpq31EbACRVAe8mc8W1T5uIaAG2AiNzFS3pakm1kmrXr1/fhdPsGbubW7ln0SrOP3YcQ8tLe+19zczSkrfprCU9BIzLsekLXT1EjnVxMG0klQC3Ad+KiBUH+j4RcQtwC0B1dXVntfSYB5asYfvuFi71QAgzGyDyFk4RcU572yStlTQ+IlZLGg+sy7FbPZD923gSsCpZbq99R20gEyzLIuIbOd6nPgmvYcCmzs+w99TU1jFpeAWnHp7zgs7MrN9Jq1tvHnBFsnwFcE+OfeYD0yVNlVRGZqDDvE7azwMuS0bgTQWmA08DSPp3MsHz2Q5qeT/wSET02lVRZ+o2NfD48o1cMnMCOjaHAAAMPklEQVQyRUUF91GYmVlepBVONwHnSloGnJu8RtIESb+GvZ//XAc8ALwI1ETEko7aJ9trgBeA3wCfiohWSZPIdCfOABZKWiTp48mxbgVGSloOfI4cIwfTdMeCeiS4eKYneTWzgUMFdJHQp1RXV0dtbW1e36OtLXjrV37H4aMr+dlVp+T1vczMeoOkBRFR3dl+niGigD3x8kZWbtnlGSHMbMBxOBWwuQvqGFpewjtmjE27FDOzXuVwKlBbG5q5//k1XHTSRMpLiztvYGbWjzicCtS8xatoamnzd5vMbEByOBWoubV1HD1uCMdMGJp2KWZmvc7hVIBeXL2NxfVbubR6MgU4zZ+ZWd45nArQ3Np6SovFRSf5u01mNjA5nApMU0sbdz9Tz7kzxjKisiztcszMUuFwKjAPv7iWzQ3N/m6TmQ1oDqcCU1Nbx7ih5bxt+ui0SzEzS43DqYCs2bqbR/+8notnTqTYk7ya2QDmcCogdz1TT1vAJTPdpWdmA5vDqUBEBHNr65k1dQRTRlWmXY6ZWaocTgWi9rXNvLJhJ5fMnJR2KWZmqXM4FYia+XVUlhVzwXHj0y7FzCx1DqcCsKOxhfueW827jp9A5aCStMsxM0udw6kA/HrxahqaWrn0ze7SMzMDh1NBqKmt4/DRlZx86PC0SzEzKwgOp5S9vH4Hta9t9iSvZmZZHE4pm1tbT3GReJ8neTUz28vhlKKW1jbuXFjPXx01mjFDy9Mux8ysYDicUvTYsvWs397oSV7NzPbjcEpRzfx6RlaWcdbRY9IuxcysoDicUrJxRyMPvbiW9540kdJi/2cwM8vm34opufuZlbS0hbv0zMxycDilICKoqa3jhMlVHDVuSNrlmJkVHIdTChbXb+XPa3dwabVnhDAzy8XhlIKa2joGlRTx7hMmpF2KmVlBcjj1sl1NrcxbtIoLjhvP0PLStMsxMytIDqde9sCSNWxvbOESd+mZmbXL4dTLDhlUwrkzxnLq1JFpl2JmVrB886Beds6MsZwzY2zaZZiZFTRfOZmZWcFJJZwkjZD0oKRlyXPOGxlJOk/SUknLJd3QlfaSbkz2Xyrpncm6wZLuk/SSpCWSbsra/0pJ6yUtSh4fz+e5m5lZ59K6croBeDgipgMPJ6/3IakYuBk4H5gBXC5pRkftk+2XAccA5wHfTY4D8F8RcTRwEnCGpPOz3u72iDgxefyoh8/VzMwOUFrhdCEwO1meDVyUY59ZwPKIWBERTcCcpF1H7S8E5kREY0S8AiwHZkVEQ0T8DiA51kLAw+XMzApUWuE0NiJWAyTPuablngjUZb2uT9Z11L6jNgBIqgLeTeaKa4+LJS2WdIekdie7k3S1pFpJtevXr+/sHM3MrJvyFk6SHpL0fI7HhZ23zhwix7o4mDaSSoDbgG9FxIpk9b3AlIg4HniI16/I3nigiFsiojoiqkePHt1JKWZm1l15G0oeEee0t03SWknjI2K1pPHAuhy71QPZVzGTgFXJcnvtO2oDcAuwLCK+kVXnxqztPwS+3MmpmZlZnqXVrTcPuCJZvgK4J8c+84HpkqZKKiMz0GFeJ+3nAZdJGiRpKjAdeBpA0r8Dw4DPZr9JEm57vAd48SDOy8zMeoAiOuspy8ObSiOBGuBQ4C/AJRGxSdIE4EcRcUGy3wXAN4Bi4McR8aWO2ifbvgB8DGgBPhsR90uaROazqJeAxqSM70TEjyT9PzKh1AJsAq6NiJe6cA7rgde6+SMYBWzoZttC43MpPP3lPMDnUqgO5lwOi4hOPxdJJZwGOkm1EVGddh09wedSePrLeYDPpVD1xrl4hggzMys4DiczMys4Dqd03JJ2AT3I51J4+st5gM+lUOX9XPyZk5mZFRxfOZmZWcFxOJmZWcFxOPUiST+WtE7S82nXcrAkTZb0O0kvJrch+UzaNXWHpHJJT0t6NjmP/5N2TQdLUrGkZyT9Ku1aDoakVyU9l9zKpjbterpLUlUyb+dLyb+X09KuqTskHZV1a6FFkrZJ+mznLbv5fv7MqfdIehuwA/hpRBybdj0HI5lZY3xELJQ0BFgAXBQRL6Rc2gGRJKAyInZIKgX+CHwmIp5KubRuk/Q5oBoYGhHvSrue7pL0KlAdEX36i6uSZgN/SL70XwYMjogtadd1MJJbEa0ETomI7k5G0CFfOfWiiHiMzCwUfV5ErI6IhcnydjLTPk3suFXhiYwdycvS5NFn/2JLZkP5a8D3JSsAkoYCbwNuhcwte/p6MCXOBl7OVzCBw8l6gKQpZG7i+Kd0K+mepBtsEZkJhB+MiD55HolvAP8ItKVdSA8I4LeSFki6Ou1iuulwYD3w30lX648kVaZdVA+4jMwdHvLG4WQHRdIhwJ1k5jHclnY93RERrRFxIplZ7GdJ6pNdrpLeBayLiAVp19JDzoiIk8ncDftTSbd4X1MCnAx8LyJOAnaS487ffUnSNfkeYG4+38fhZN2WfEZzJ/CLiLgr7XoOVtLd8nvgvJRL6a4zgPckn9XMAc6S9PN0S+q+iFiVPK8D7iZzd+y+ph6oz7oav4NMWPVl5wMLI2JtPt/E4WTdkgwkuBV4MSK+lnY93SVpdHJ3ZCRVAOeQmb2+z4mIGyNiUkRMIdPt8khEfCjlsrpFUmUy0IakG+wdQJ8b5RoRa4A6SUclq84G+tSgoRwuJ89depDHmw3aG0m6DTgTGCWpHvjXiLg13aq67Qzgw8Bzyec1AP8UEb9OsabuGA/MTkYfFQE1EdGnh2D3E2OBuzN/A1EC/DIifpNuSd32aeAXSXfYCuCjKdfTbZIGA+cCn8z7e3kouZmZFRp365mZWcFxOJmZWcFxOJmZWcFxOJmZWcFxOJmZWcFxOJnlICkkfTXr9d9L+mIPHXtH53t1+9i/l1R9kMeY0t2Z8yVdKWnCwby/GTiczNrTCLxP0qg03lxSX/0O4pXAAYVTHz5XyyOHk1luLcAtwN/tv0HSYZIelrQ4eT40Wf8TSd9L7nO1QtLbk3t4vSjpJ/sd46uSFibtRyfrfi/pPyQ9Cnwmmb3iTknzk8cZOWqpkDQnqeV2oCJr2zskPZm8z9xkHsT92x8h6aHkflYLJU3bb/uVkr6T9fpXks5MJsv9iaTnk3su/Z2k95O5Vccvkvv9VEiaKenRZPLWB5JbreQ610uSYz0r6bGu/2ey/srhZNa+m4EPShq23/rvkLkn1/HAL4BvZW0bDpxFJtTuBb4OHAMcJ+nEZJ9KMnOTnQw8CvxrVvuqiHh7RHwV+Cbw9Yh4M3AxuW+DcS3QkNTyJWAmQHLF98/AOcn71AKfy9H+F8DNEXECcDqwupOfyR4nAhMj4tiIOA7474i4I3mfDyYT6bYA3wbeHxEzgR8nNeY6138B3pnU8Z4u1mD9mC+nzdoREdsk/RS4HtiVtek04H3J8s+Ar2RtuzciQtJzwNqIeA5A0hJgCrCIzO0sbk/2/zmQPWnu7VnL5wAzkil8AIZKGpLcP2uPt5GEY0QslrQ4WX8qMAN4PGlfBjyZfX7J3HUTI+LupP3uZH0HP5W9VgCHS/o2cB/w2xz7HAUcCzyYHLOYfcMv+1wfB34iqYZ9fx42QDmczDr2DWAh8N8d7JM9B1hj8tyWtbzndXv/3rLb78xaLgJOi4hddCzXHGQic2+qyzto15UUamHfHpZygIjYLOkE4J3Ap4BLgY/lOP6SiGjvtuR7zzUirpF0CpkbJS6SdGJEbOxCfdZPuVvPrAMRsQmoAa7KWv0EmVm/AT5I5tbuB6IIeH+y/DcdtP8tcN2eF1ndgtkeS2oguQ/V8cn6p4AzJB2RbBss6cjshsn9t+olXZTsMyiZ2DPbq8CJkookTSa5bUXSbVgUEXcC/5vXbwOxHRiSLC8FRks6LWlTKumYXCcqaVpE/Cki/gXYAExu52diA4SvnMw691WyQoJMN9+PJf0DmbucHugs0zuBYyQtALYCH2hnv+uBm5OuuhIyQXTNfvt8j8xdVheT6TJ8GiAi1ku6ErhN0qBk338G/rxf+w8DP5D0b0AzcAn73kX3ceAV4Dkyt6xYmKyfmLzvnj9wb0yefwJ8X9IuMt2f7we+lXxuV0LmSnRJjnP9T0nTyVxtPQw8287PxAYIz0puZmYFx916ZmZWcBxOZmZWcBxOZmZWcBxOZmZWcBxOZmZWcBxOZmZWcBxOZmZWcP4/CmzOZ6xennYAAAAASUVORK5CYII=\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "K_clusters = range(1, n_pumps)\n", "kmeans = [KMeans(n_clusters=i) for i in K_clusters]\n", "Y_axis = data_death[['x']]\n", "X_axis = data_death[['y']]\n", "score = [kmeans[i].fit(Y_axis).score(Y_axis) for i in range(len(kmeans))]\n", "plt.plot(K_clusters, score)\n", "plt.xlabel('Nombre de clusters')\n", "plt.ylabel('Score')\n", "plt.title('Elbow Curve')\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "La courbe nous montre que le nombre de **K** optimal est **3**." ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [], "source": [ "kmeans = KMeans(n_clusters = 3, init ='k-means++')\n", "kmeans.fit(data_death[data_death.columns[1:3]])\n", "data_death['cluster_label'] = kmeans.fit_predict(data_death[data_death.columns[1:3]])\n", "centers = kmeans.cluster_centers_\n", "labels = kmeans.predict(data_death[data_death.columns[1:3]])" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "data_death.plot.scatter(x = 'x', y = 'y', c=labels, s=20, \n", " ylim=[data_pumps['y'].min()-0.001, data_pumps['y'].max()-0.001],\n", " xlim=[data_pumps['x'].min()+0.0015, data_pumps['x'].max()+0.001], cmap='viridis')\n", "plt.scatter(centers[:, 0], centers[:, 1], c='red', s=100, alpha=0.8)\n", "plt.show()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "On récupère le cluster qui contient le plus de cas de décès." ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Cluster avec le plus grand nombre de morts : 0\n" ] } ], "source": [ "cluster_by_death = data_death.groupby('cluster_label').count()['d_count']\n", "max_cluster = cluster_by_death.idxmax()\n", "print('Cluster avec le plus grand nombre de morts : {}'.format(max_cluster))" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "cluster_label\n", "0 202\n", "1 137\n", "2 150\n", "Name: d_count, dtype: int64" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "cluster_by_death" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "On peut placer les clusters sur la map pour en avoir une meilleure représentation et vérifier que le cluster trouvé se trouve près de la pompe de Broad St." ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
Make this Notebook Trusted to load map: File -> Trust Notebook
" ], "text/plain": [ "" ] }, "execution_count": 21, "metadata": {}, "output_type": "execute_result" } ], "source": [ "cluster_map = death_pump_map\n", "for p in range(0, len(centers)):\n", " folium.Marker(centers[p],\n", " popup='Cluster : {}'.format(p),\n", " icon=folium.Icon(color='green')).add_to(death_pump_map)\n", "cluster_map" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "On vérifie notre hypothèse visuellement. On peut aussi calculer **la distance euclidienne** entre le cluster 0 et les pompes pour vérifier qu'il est au plus près de la pompe de Broad St." ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [], "source": [ "c_0_coordinates = centers[max_cluster]" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [], "source": [ "euclidean_distances = []\n", "for i in pump_coordinates:\n", " euclidean_distances.append(distance.euclidean(i, c_0_coordinates))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "L'indice de la distance minimale nous donne l'indice de la pompe au centre de l'épidémie." ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [], "source": [ "pump_idx = euclidean_distances.index(min(euclidean_distances))" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "name Broad St.\n", "x 51.5133\n", "y -0.136668\n", "Name: 0, dtype: object" ] }, "execution_count": 25, "metadata": {}, "output_type": "execute_result" } ], "source": [ "data_pumps.iloc[pump_idx]" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "**On prouve donc par clustering que la pompe de Broad St. est au centre de l'épidémie.**" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.4" } }, "nbformat": 4, "nbformat_minor": 4 }