{"id":6558,"date":"2020-07-29T09:00:53","date_gmt":"2020-07-29T09:00:53","guid":{"rendered":"https:\/\/datascientest.com\/?p=6558"},"modified":"2023-11-09T14:12:50","modified_gmt":"2023-11-09T14:12:50","slug":"machine-learning-clustering-dbscan","status":"publish","type":"post","link":"https:\/\/datascientest.com\/machine-learning-clustering-dbscan","title":{"rendered":"Machine Learning &amp; Clustering: Focus sur l&rsquo;Algorithme DBSCAN"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"6558\" class=\"elementor elementor-6558\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-bc65502 e-flex e-con-boxed e-con\" data-id=\"bc65502\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-471031b elementor-widget elementor-widget-text-editor\" data-id=\"471031b\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#69727d;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#69727d;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<h2>Le clustering est une discipline particuli\u00e8re du Machine Learning ayant pour objectif de s\u00e9parer vos donn\u00e9es en groupes homog\u00e8nes ayant des caract\u00e9ristiques communes. C\u2019est un domaine tr\u00e8s appr\u00e9ci\u00e9 en marketing, par exemple, o\u00f9 l\u2019on cherche souvent \u00e0 segmenter les bases clients pour d\u00e9tecter des comportements particuliers.<\/h2>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1ef7f91 elementor-widget elementor-widget-text-editor\" data-id=\"1ef7f91\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Dans de pr\u00e9c\u00e9dents articles, nous avions pr\u00e9sent\u00e9 deux algorithmes de clustering :\u00a0<a href=\"https:\/\/datascientest.com\/clustering-focus-sur-lalgorithme-des-k-moyennes\" target=\"_blank\" rel=\"noopener\">celui des K-moyennes\/ K-means<\/a>\u00a0et l\u2019<a href=\"https:\/\/datascientest.com\/machine-learning-clustering-focus-sur-lalgorithme-cah\" target=\"_blank\" rel=\"noopener\">algorithme CAH-classification ascendante hi\u00e9rarchique<\/a>.\u00a0<\/p><p>Il nous semblait essentiel de poursuivre cet apprentissage avec le DBSCAN.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-5ad6369 e-flex e-con-boxed e-con\" data-id=\"5ad6369\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-4dde8f6 elementor-widget elementor-widget-heading\" data-id=\"4dde8f6\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-heading-title{padding:0;margin:0;line-height:1}.elementor-widget-heading .elementor-heading-title[class*=elementor-size-]>a{color:inherit;font-size:inherit;line-height:inherit}.elementor-widget-heading .elementor-heading-title.elementor-size-small{font-size:15px}.elementor-widget-heading .elementor-heading-title.elementor-size-medium{font-size:19px}.elementor-widget-heading .elementor-heading-title.elementor-size-large{font-size:29px}.elementor-widget-heading .elementor-heading-title.elementor-size-xl{font-size:39px}.elementor-widget-heading .elementor-heading-title.elementor-size-xxl{font-size:59px}<\/style><h3 class=\"elementor-heading-title elementor-size-default\">L\u2019algorithme DBSCAN<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-edad377 elementor-widget elementor-widget-text-editor\" data-id=\"edad377\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Le <b>DBSCAN<\/b> est un <b>algorithme non supervis\u00e9<\/b> tr\u00e8s connu en mati\u00e8re de <b>Clustering<\/b>. Il a \u00e9t\u00e9 propos\u00e9 1996 par Martin Ester, Hans-Peter Kriegel, J\u00f6rg Sander et Xiawei Xu.\u00a0<\/p><p>Dans cet article, nous allons d\u00e9tailler son fonctionnement et <b>comment l\u2019impl\u00e9menter en Python<\/b> \u00e0 l\u2019aide de librairies tel que\u00a0<a href=\"https:\/\/scikit-learn.org\/stable\/\" target=\"_blank\" rel=\"noopener\">Scikit-Learn<\/a>.\u00a0<span style=\"font-size: 16px;\">Ses champs d\u2019application sont divers : analyse cartographique, analyse de donn\u00e9e, segmenter une image\u2026.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-9d57aa5 elementor-widget elementor-widget-heading\" data-id=\"9d57aa5\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\">Le principe<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1eaff07 elementor-widget elementor-widget-text-editor\" data-id=\"1eaff07\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>\u00c9tant donn\u00e9s des points et un entier k, l&rsquo;algorithme vise \u00e0 <strong>diviser les points en k groupes<\/strong>, appel\u00e9s <strong>clusters<\/strong>, homog\u00e8nes et compacts. Regardons l\u2019exemple ci-dessous :<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d4f2efa elementor-widget elementor-widget-image\" data-id=\"d4f2efa\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-image{text-align:center}.elementor-widget-image a{display:inline-block}.elementor-widget-image a img[src$=\".svg\"]{width:48px}.elementor-widget-image img{vertical-align:middle;display:inline-block}<\/style>\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"353\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN1.png\" class=\"attachment-large size-large wp-image-6560\" alt=\"dbscan\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN1.png 883w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN1-300x133.png 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN1-768x339.png 768w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1b4c57a elementor-widget elementor-widget-text-editor\" data-id=\"1b4c57a\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Le DBSCAN est un algorithme simple qui d\u00e9finit des clusters en utilisant l\u2019<strong>estimation de la densit\u00e9 locale<\/strong>. On peut le diviser en 4 \u00e9tapes\u00a0:<\/p><ul><li>Pour chaque observation on regarde le nombre de points \u00e0 au plus une distance \u03b5 de celle-ci. On appelle cette zone le <strong>\u03b5-voisinage de l\u2019observation<\/strong>.<\/li><li>Si une observation compte au moins un certain nombre de voisins y compris elle-m\u00eame, elle est consid\u00e9r\u00e9e comme une <strong>observation c\u0153ur<\/strong>. On a alors d\u00e9cel\u00e9 une <strong>observation \u00e0 haute densit\u00e9<\/strong>.<\/li><li>Toutes les observations au voisinage d\u2019une observation c\u0153ur appartiennent au m\u00eame cluster. Il peut y avoir des observations c\u0153ur proche les unes des autres. <strong>Par cons\u00e9quent de proche en proche on obtient une longue s\u00e9quence d\u2019observations c\u0153ur qui constitue un unique cluster<\/strong>.<\/li><li>Toute observation qui n\u2019est pas une observation c\u0153ur et qui ne comporte pas d\u2019observation c\u0153ur dans son voisinage est consid\u00e9r\u00e9e comme une <strong>anomalie<\/strong>.<\/li><\/ul><p>Vous avez donc besoin de d\u00e9finir deux informations avant d\u2019utiliser le DBSCAN\u00a0:<\/p><blockquote><p>Quelle distance \u03b5 pour d\u00e9terminer pour chaque observation le <strong>\u03b5<\/strong>-voisinage\u00a0? Quel est le <strong>nombre minimal de voisins n\u00e9cessaire<\/strong> pour consid\u00e9rer qu\u2019une observation est une <strong>observation c\u0153ur<\/strong>\u00a0?<\/p><\/blockquote><p>Ces deux informations sont renseign\u00e9es librement par l\u2019utilisateur. Contrairement \u00e0 l\u2019algorithme des k-moyennes ou la classification ascendante hi\u00e9rarchique, il n\u2019y a <strong>pas besoin de d\u00e9finir en amont le nombre de clusters<\/strong> ce qui rend l\u2019algorithme moins rigide.\u00a0<\/p><p>Un autre avantage de DBSCAN est qu\u2019il permet aussi de <strong>g\u00e9rer les valeurs aberrantes ou anomalies<\/strong>. Vous remarquerez dans la figure ci-dessus que l\u2019algorithme a d\u00e9termin\u00e9 3 clusters principaux\u00a0: le bleu, le vert et le jaune. Les points color\u00e9s en violet constituent des anomalies d\u00e9tect\u00e9es par le DBSCAN. \u00c9videmment suivant la valeur de \u03b5 et le nombre de voisins minimal le partitionnement peut varier.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f04dbce elementor-widget elementor-widget-heading\" data-id=\"f04dbce\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-default\">Notion de distance et choix du \u03b5<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f1d9c2c elementor-widget elementor-widget-text-editor\" data-id=\"f1d9c2c\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Dans cet algorithme deux points sont\u00a0 cl\u00e9s :\u00a0<\/span><\/p><blockquote><p><span style=\"font-weight: 400;\">Quelle est la m\u00e9trique utilis\u00e9e pour \u00e9valuer la distance entre une observation et ses voisins ? quel est le <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\"> id\u00e9al\u00a0?<\/span><\/p><\/blockquote><p><span style=\"font-weight: 400;\">Dans le DBSCAN on utilise g\u00e9n\u00e9ralement la distance euclidienne, soient p = (p1,\u2026.,pn) et q = (q1,\u2026.,qn)\u00a0: <\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-81bbfa7 elementor-widget elementor-widget-image\" data-id=\"81bbfa7\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"79\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan2.png\" class=\"attachment-large size-large wp-image-6563\" alt=\"distance euclidienne\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan2.png 833w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan2-300x30.png 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan2-768x76.png 768w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c35df67 elementor-widget elementor-widget-text-editor\" data-id=\"c35df67\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">\u00c0 chaque observation, pour compter le nombre de voisins \u00e0 au plus une distance <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\"> , on calcule la distance euclidienne entre le voisin et l\u2019observation et v\u00e9rifie si c\u2019est inf\u00e9rieur \u00e0 <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\">.<\/span><\/p><p><span style=\"font-weight: 400;\">Reste maintenant \u00e0 savoir comment choisir le bon epsilon. Supposons que dans notre exemple nous choisissons de tester l\u2019algorithme avec des valeurs diff\u00e9rentes de <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\"> . Voici le r\u00e9sultat\u00a0: <\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-4ffc779 elementor-widget elementor-widget-image\" data-id=\"4ffc779\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"800\" height=\"288\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan3-1024x368.png\" class=\"attachment-large size-large wp-image-6564\" alt=\"dbscan\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan3-1024x368.png 1024w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan3-300x108.png 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan3-768x276.png 768w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/dbscan3.png 1078w\" sizes=\"(max-width: 800px) 100vw, 800px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-14311fb elementor-widget elementor-widget-text-editor\" data-id=\"14311fb\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Dans les trois exemples le nombre de voisins minimal est toujours fix\u00e9 \u00e0 5.\u00a0<\/span><\/p><blockquote><p><span style=\"font-weight: 400;\">Si <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\"> est trop petit le <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\">-voisinage est <\/span><b>trop faible <\/b><span style=\"font-weight: 400;\">et <\/span><b>toutes les observations <\/b><span style=\"font-weight: 400;\">du jeu de donn\u00e9es sont consid\u00e9r\u00e9es comme des <\/span><b>anomalies<\/b><span style=\"font-weight: 400;\">.\u00a0<\/span><\/p><\/blockquote><p><span style=\"font-weight: 400;\">C\u2019est le cas de la figure de gauche eps = 0.05.\u00a0<\/span><\/p><blockquote><p><span style=\"font-weight: 400;\">A contrario si <\/span><b>epsilon est trop grand<\/b><span style=\"font-weight: 400;\"> chaque observation contient dans son <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\">-voisinage toutes les autres observations du jeu de donn\u00e9es. Par cons\u00e9quent nous n\u2019obtenons qu\u2019<\/span><b>un unique cluster<\/b><span style=\"font-weight: 400;\">. Il est donc tr\u00e8s important de <\/span><b>bien calibrer le \u03b5 <\/b><span style=\"font-weight: 400;\">pour obtenir un partitionnement de qualit\u00e9.\u00a0<\/span><\/p><\/blockquote><p><span style=\"font-weight: 400;\">Une m\u00e9thode simple pour optimiser le <\/span><b>\u03b5 <\/b><span style=\"font-weight: 400;\">consiste \u00e0 regarder pour chaque observation \u00e0 quelle distance se situe son voisin le plus proche. Ensuite il suffit de fixer un <\/span><b>\u03b5 <\/b><span style=\"font-weight: 400;\">tel qu&rsquo;une part \u00ab\u00a0suffisamment grande\u00a0\u00bb des observations aient une distance \u00e0 son plus proche voisin inf\u00e9rieure \u00e0<\/span> <b>\u03b5<\/b><span style=\"font-weight: 400;\">. Par \u00ab\u00a0suffisamment grande\u00a0\u00bb on entend 90-95% des observations qui doivent avoir au moins un voisin dans leur <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\">-voisinage.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-bc97e26 e-flex e-con-boxed e-con\" data-id=\"bc97e26\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-c61f5ab elementor-widget elementor-widget-heading\" data-id=\"c61f5ab\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Comment impl\u00e9menter le DBSCAN avec Python ?<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e6b2d63 elementor-widget elementor-widget-text-editor\" data-id=\"e6b2d63\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">L\u2019entra\u00eenement du DBSCAN est facilit\u00e9 avec la <\/span><b>librairie Scikit-Learn<\/b><span style=\"font-weight: 400;\">. Notre jeu de donn\u00e9es qui a servi d\u2019exemple se cr\u00e9e facilement avec la fonctionalit\u00e9 make_blobs de Scikit-Learn. Voici comment l\u2019impl\u00e9menter en code\u00a0:<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-44fff94 elementor-widget elementor-widget-shortcode\" data-id=\"44fff94\" data-element_type=\"widget\" data-widget_type=\"shortcode.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-shortcode\"><script src=\"https:\/\/gist.github.com\/Thibault-venet\/ddb1f1b944aafb2db333c0016963df4c.js\"><\/script><\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1cae59e elementor-widget elementor-widget-text-editor\" data-id=\"1cae59e\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<blockquote><p>Maintenant que nous avons notre jeu de donn\u00e9es nous allons chercher \u00e0 d\u00e9terminer le<strong> \u03b5 optimal<\/strong> pour obtenir un <strong>meilleur partitionnement<\/strong> de notre jeu de donn\u00e9es.\u00a0<\/p><\/blockquote><p>Scikit-Learn met \u00e0 disposition une\u00a0<a href=\"https:\/\/scikit-learn.org\/stable\/modules\/generated\/sklearn.neighbors.NearestNeighbors.html\">classe NearestNeighbors<\/a>\u00a0qui permet de d\u00e9terminer les voisins les plus proches de chaque observation ainsi que les distances. Voici comment l\u2019impl\u00e9menter en code\u00a0:<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-2df8bdc elementor-widget elementor-widget-shortcode\" data-id=\"2df8bdc\" data-element_type=\"widget\" data-widget_type=\"shortcode.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-shortcode\"><script src=\"https:\/\/gist.github.com\/Thibault-venet\/ff05061e12f8232715aaafc3b9ab0377.js\"><\/script><\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-f28e7f1 elementor-widget elementor-widget-text-editor\" data-id=\"f28e7f1\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Voici le r\u00e9sultat obtenu :<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-45c5bb7 elementor-widget elementor-widget-image\" data-id=\"45c5bb7\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"460\" height=\"309\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/nearestneighbors.png\" class=\"attachment-large size-large wp-image-6565\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/nearestneighbors.png 460w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/nearestneighbors-300x202.png 300w\" sizes=\"(max-width: 460px) 100vw, 460px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c0b3b85 elementor-widget elementor-widget-text-editor\" data-id=\"c0b3b85\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Comme nous l\u2019avons vu pr\u00e9c\u00e9demment nous allons choisir un <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\"> de tel sorte que 90% des observations aient une distance au proche voisin inf\u00e9rieure \u00e0 <\/span><b>\u03b5<\/b><span style=\"font-weight: 400;\">. Dans notre exemple 0.2 semble convenir.<\/span><\/p><p><span style=\"font-weight: 400;\">Maintenant que nous avons optimis\u00e9 notre <\/span><b>\u03b5 <\/b><span style=\"font-weight: 400;\">nous pouvons effectuer notre partitionnement en utilisant DBSCAN qui est d\u00e9j\u00e0 impl\u00e9ment\u00e9 dans Scikit-Learn.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-3421851 elementor-widget elementor-widget-shortcode\" data-id=\"3421851\" data-element_type=\"widget\" data-widget_type=\"shortcode.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-shortcode\"><script src=\"https:\/\/gist.github.com\/Thibault-venet\/b5cfc21f887ae9afaf361d1b827f66c9.js\"><\/script><\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ae3ea50 elementor-widget elementor-widget-text-editor\" data-id=\"ae3ea50\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Voici le r\u00e9sultat obtenu :<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d7c2d2b elementor-widget elementor-widget-image\" data-id=\"d7c2d2b\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"453\" height=\"315\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN4.png\" class=\"attachment-large size-large wp-image-6566\" alt=\"DBSCAN\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN4.png 453w, https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN4-300x209.png 300w\" sizes=\"(max-width: 453px) 100vw, 453px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-95fbe7c elementor-widget elementor-widget-text-editor\" data-id=\"95fbe7c\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Les arguments eps et min_samples permettent de fixer les param\u00e8tres de l\u2019algorithme \u00e0 savoir la distance \u03b5 et le nombre minimal de voisins pour \u00eatre consid\u00e9r\u00e9 comme une observation c\u0153ur.<\/p><p>L\u2019algorithme a donc d\u00e9tect\u00e9 3 principaux clusters et quelques anomalies. Chacune des observations est \u00e9tiquet\u00e9e avec l\u2019\u00e9tiquette du cluster correspondant (par exemple 0,1,2). Les anomalies sont \u00e9tiquet\u00e9es en -1 et forment donc une classe \u00e0 part en plus des 3 clusters d\u00e9tect\u00e9s.\u00a0<\/p><p>L\u2019utilisation d\u2019un algorithme de clustering tel que le DBSCAN n\u2019est pas ais\u00e9e suivant les donn\u00e9es dont vous disposez. Il faudra souvent les retravailler.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-098a25a e-flex e-con-boxed e-con\" data-id=\"098a25a\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-ea1651d elementor-widget elementor-widget-text-editor\" data-id=\"ea1651d\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>Cette s\u00e9rie sur le Clustering vous a plu ? Les <strong>formations DataScientest<\/strong> vous apprendrons \u00e0 ma\u00eetriser tous les principaux algorithmes de clustering ainsi qu\u2019\u00e0 les interpr\u00e9ter sur des jeux de donn\u00e9es.<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-61a4c37 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"61a4c37\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formations-data\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">D\u00e9couvrir les formations DataScientest<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c613fe9 elementor-widget elementor-widget-text-editor\" data-id=\"c613fe9\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p>D\u00e9couvrez les\u00a0<a href=\"https:\/\/datascientest.com\/#dateformations\" target=\"_blank\" rel=\"noopener\">prochaines dates de lancement\u00a0<\/a>rapidement !<\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Le clustering est une discipline particuli\u00e8re du Machine Learning ayant pour objectif de s\u00e9parer vos donn\u00e9es en groupes homog\u00e8nes ayant des caract\u00e9ristiques communes. C\u2019est un domaine tr\u00e8s appr\u00e9ci\u00e9 en marketing, par exemple, o\u00f9 l\u2019on cherche souvent \u00e0 segmenter les bases clients pour d\u00e9tecter des comportements particuliers. Dans de pr\u00e9c\u00e9dents articles, nous avions pr\u00e9sent\u00e9 deux algorithmes [&hellip;]<\/p>\n","protected":false},"author":78,"featured_media":6566,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"content-type":"","inline_featured_image":false},"categories":[2497],"tags":[1249],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.7 (Yoast SEO v21.7) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>Machine Learning &amp; Clustering: Focus sur l&#039;Algorithme DBSCAN<\/title>\n<meta name=\"description\" content=\"DBSCAN est un algorithme non supervis\u00e9 populaire en Clustering, discipline essentielle du Machine Learning. D\u00e9couvrez son fonctionnement !\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/datascientest.com\/machine-learning-clustering-dbscan\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Machine Learning &amp; Clustering: Focus sur l&#039;Algorithme DBSCAN\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/machine-learning-clustering-dbscan\" \/>\n<meta property=\"og:site_name\" content=\"Formation Data Science | DataScientest.com\" \/>\n<meta property=\"article:published_time\" content=\"2020-07-29T09:00:53+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-11-09T14:12:50+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN4.png\" \/>\n\t<meta property=\"og:image:width\" content=\"453\" \/>\n\t<meta property=\"og:image:height\" content=\"315\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"J\u00e9r\u00e9my Robert\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"J\u00e9r\u00e9my Robert\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"6 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/datascientest.com\/machine-learning-clustering-dbscan\",\"url\":\"https:\/\/datascientest.com\/machine-learning-clustering-dbscan\",\"name\":\"Machine Learning &amp; Clustering: Focus sur l'Algorithme DBSCAN\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/#website\"},\"datePublished\":\"2020-07-29T09:00:53+00:00\",\"dateModified\":\"2023-11-09T14:12:50+00:00\",\"author\":{\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/7af2361c48e24e79a8aa2360e6e74925\"},\"description\":\"DBSCAN est un algorithme non supervis\u00e9 populaire en Clustering, discipline essentielle du Machine Learning. D\u00e9couvrez son fonctionnement !\",\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/machine-learning-clustering-dbscan\"]}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/datascientest.com\/#website\",\"url\":\"https:\/\/datascientest.com\/\",\"name\":\"Formation Data Science | DataScientest.com\",\"description\":\"Leader de la formation en Data Science\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/datascientest.com\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/7af2361c48e24e79a8aa2360e6e74925\",\"name\":\"J\u00e9r\u00e9my Robert\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/jeremyrobert.jpeg\",\"contentUrl\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/jeremyrobert.jpeg\",\"caption\":\"J\u00e9r\u00e9my Robert\"},\"description\":\"J\u00e9r\u00e9my Robert est actuellement Directeur p\u00e9dagogique B2B chez DataScientest.com. Avant de rejoindre DataScientest, il a travaill\u00e9 en tant que Consultant Business Intelligence chez DEAPDATA, o\u00f9 il a g\u00e9r\u00e9 des projets dans les domaines de l'assurance, bancaire et industrie. Il a \u00e9galement \u00e9t\u00e9 Co-fondateur et pr\u00e9sident de Tripzer. En tant que consultant ind\u00e9pendant, J\u00e9r\u00e9my a propos\u00e9 des formations sur MicroStrategy pour divers niveaux professionnels. Il est dipl\u00f4m\u00e9 de l'Universit\u00e9 Paris Cit\u00e9 avec un Master MIAGE, combinant comp\u00e9tences en informatique et gestion.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/jeremyrobert1\/\",\"Data Science\"],\"url\":\"https:\/\/datascientest.com\/author\/jeremyrobert\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Machine Learning &amp; Clustering: Focus sur l'Algorithme DBSCAN","description":"DBSCAN est un algorithme non supervis\u00e9 populaire en Clustering, discipline essentielle du Machine Learning. D\u00e9couvrez son fonctionnement !","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/datascientest.com\/machine-learning-clustering-dbscan","og_locale":"fr_FR","og_type":"article","og_title":"Machine Learning &amp; Clustering: Focus sur l'Algorithme DBSCAN","og_url":"https:\/\/datascientest.com\/machine-learning-clustering-dbscan","og_site_name":"Formation Data Science | DataScientest.com","article_published_time":"2020-07-29T09:00:53+00:00","article_modified_time":"2023-11-09T14:12:50+00:00","og_image":[{"width":453,"height":315,"url":"https:\/\/datascientest.com\/wp-content\/uploads\/2020\/07\/DBSCAN4.png","type":"image\/png"}],"author":"J\u00e9r\u00e9my Robert","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"J\u00e9r\u00e9my Robert","Dur\u00e9e de lecture estim\u00e9e":"6 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/datascientest.com\/machine-learning-clustering-dbscan","url":"https:\/\/datascientest.com\/machine-learning-clustering-dbscan","name":"Machine Learning &amp; Clustering: Focus sur l'Algorithme DBSCAN","isPartOf":{"@id":"https:\/\/datascientest.com\/#website"},"datePublished":"2020-07-29T09:00:53+00:00","dateModified":"2023-11-09T14:12:50+00:00","author":{"@id":"https:\/\/datascientest.com\/#\/schema\/person\/7af2361c48e24e79a8aa2360e6e74925"},"description":"DBSCAN est un algorithme non supervis\u00e9 populaire en Clustering, discipline essentielle du Machine Learning. D\u00e9couvrez son fonctionnement !","inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/machine-learning-clustering-dbscan"]}]},{"@type":"WebSite","@id":"https:\/\/datascientest.com\/#website","url":"https:\/\/datascientest.com\/","name":"Formation Data Science | DataScientest.com","description":"Leader de la formation en Data Science","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/datascientest.com\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Person","@id":"https:\/\/datascientest.com\/#\/schema\/person\/7af2361c48e24e79a8aa2360e6e74925","name":"J\u00e9r\u00e9my Robert","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/datascientest.com\/#\/schema\/person\/image\/","url":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/jeremyrobert.jpeg","contentUrl":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/jeremyrobert.jpeg","caption":"J\u00e9r\u00e9my Robert"},"description":"J\u00e9r\u00e9my Robert est actuellement Directeur p\u00e9dagogique B2B chez DataScientest.com. Avant de rejoindre DataScientest, il a travaill\u00e9 en tant que Consultant Business Intelligence chez DEAPDATA, o\u00f9 il a g\u00e9r\u00e9 des projets dans les domaines de l'assurance, bancaire et industrie. Il a \u00e9galement \u00e9t\u00e9 Co-fondateur et pr\u00e9sident de Tripzer. En tant que consultant ind\u00e9pendant, J\u00e9r\u00e9my a propos\u00e9 des formations sur MicroStrategy pour divers niveaux professionnels. Il est dipl\u00f4m\u00e9 de l'Universit\u00e9 Paris Cit\u00e9 avec un Master MIAGE, combinant comp\u00e9tences en informatique et gestion.","sameAs":["https:\/\/www.linkedin.com\/in\/jeremyrobert1\/","Data Science"],"url":"https:\/\/datascientest.com\/author\/jeremyrobert"}]}},"_links":{"self":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/6558"}],"collection":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/users\/78"}],"replies":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/comments?post=6558"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/6558\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media\/6566"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media?parent=6558"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/categories?post=6558"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/tags?post=6558"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}