{"id":184112,"date":"2022-12-20T16:32:46","date_gmt":"2022-12-20T16:32:46","guid":{"rendered":"https:\/\/datascientest.com\/?p=184112"},"modified":"2023-11-09T13:55:10","modified_gmt":"2023-11-09T13:55:10","slug":"underfitting-tout-savoir","status":"publish","type":"post","link":"https:\/\/datascientest.com\/underfitting-tout-savoir","title":{"rendered":"Underfitting ou le sous-ajustement en Machine Learning"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"184112\" class=\"elementor elementor-184112\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-01b7de9 e-flex e-con-boxed e-con\" data-id=\"01b7de9\" 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-0b6c69a elementor-widget elementor-widget-heading\" data-id=\"0b6c69a\" 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><h2 class=\"elementor-heading-title elementor-size-default\">La d\u00e9finition du bon mod\u00e8le est primordiale pour r\u00e9aliser des pr\u00e9dictions pertinentes en Machine Learning. Mais un mauvais ajustement des donn\u00e9es d\u2019apprentissage peut affecter la performance des analyses pr\u00e9dictives. C\u2019est justement ce qui se passe avec l\u2019underfitting. Alors de quoi s\u2019agit-il ? Et comment l\u2019\u00e9viter ? Les r\u00e9ponses sont ici.<\/h2>\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-74ed47d e-flex e-con-boxed e-con\" data-id=\"74ed47d\" 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-9dca931 elementor-widget elementor-widget-heading\" data-id=\"9dca931\" 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\">Qu\u2019est-ce que l\u2019underfitting ?<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-384bee3 elementor-widget elementor-widget-text-editor\" data-id=\"384bee3\" 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<p>L\u2019underfitting est une <strong>notion incontournable du Machine Learning<\/strong>, puisqu\u2019il peut \u00eatre responsable d\u2019une mauvaise performance de l\u2019<a href=\"https:\/\/datascientest.com\/apprentissage-supervise\">apprentissage supervis\u00e9<\/a> \u00e0 partir des donn\u00e9es. Alors avant de voir plus en d\u00e9tail cette notion, il convient de rappeler quelques <strong>\u00e9l\u00e9ments essentiels du fonctionnement du Machine Learning<\/strong>.<\/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-a7b8844 e-flex e-con-boxed e-con\" data-id=\"a7b8844\" 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-82fa2e6 elementor-widget elementor-widget-heading\" data-id=\"82fa2e6\" 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\">Rappel sur le fonctionnement du Machine Learning<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d9b2a9b elementor-widget elementor-widget-text-editor\" data-id=\"d9b2a9b\" 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 Machine Learning a pour objectif de <strong>pr\u00e9dire un pattern \u00e0 partir de mod\u00e8les d\u2019apprentissage sur des donn\u00e9es encore inconnues<\/strong>.<\/p><p>Pour cela, l\u2019apprentissage supervis\u00e9 repose sur deux id\u00e9es ma\u00eetresses. \u00c0 savoir :<\/p><ul><li><strong>L\u2019approximation<\/strong> : il s\u2019agit d\u2019entra\u00eener des donn\u00e9es d\u2019apprentissage (ou training set) pour obtenir un mod\u00e8le pr\u00e9sentant un taux d\u2019erreur le plus faible possible.<\/li><li><strong>La g\u00e9n\u00e9ralisation<\/strong> : suite \u00e0 cet entra\u00eenement, les Data Scientists utilisent des donn\u00e9es de validation pour tester les diff\u00e9rents mod\u00e8les. Cela permet de concevoir un mod\u00e8le capable de se g\u00e9n\u00e9raliser \u00e0 de nouvelles donn\u00e9es jamais vues auparavant.<\/li><\/ul><p>L\u2019id\u00e9e globale est d\u2019obtenir des niveaux d\u2019erreur en apprentissage et en validation similaire.<\/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-94b2657 elementor-widget elementor-widget-image\" data-id=\"94b2657\" 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=\"802\" height=\"539\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/machine-learning-underfitting.jpg\" class=\"attachment-full size-full wp-image-184127\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/machine-learning-underfitting.jpg 802w, https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/machine-learning-underfitting-300x202.jpg 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/machine-learning-underfitting-768x516.jpg 768w\" sizes=\"(max-width: 802px) 100vw, 802px\" \/>\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\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-a6ced7f e-flex e-con-boxed e-con\" data-id=\"a6ced7f\" 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-de2ad82 elementor-widget elementor-widget-heading\" data-id=\"de2ad82\" 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\">L\u2019underfitting ou le sous-apprentissage<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-af71ea4 elementor-widget elementor-widget-text-editor\" data-id=\"af71ea4\" 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>L\u2019underfitting peut \u00eatre traduit par le sous-ajustement ou le sous-apprentissage. Dans ce cas, on dit aussi que <strong>le mod\u00e8le souffre d\u2019un grand Bias<\/strong> (ou <a href=\"https:\/\/datascientest.com\/4-types-de-biais-statistique-a-eviter-dans-vos-analyses\">biais<\/a>).<\/p><p>Cela survient lorsque l\u2019on nomme uniquement des variables pertinentes en lien avec le probl\u00e8me ou lorsque l\u2019on force le mod\u00e8le \u00e0 arr\u00eater d\u2019apprendre pr\u00e9matur\u00e9ment. D\u00e8s lors, <strong>le mod\u00e8le cr\u00e9\u00e9 est trop simpliste<\/strong> et risque de passer \u00e0 c\u00f4t\u00e9 de la tendance g\u00e9n\u00e9rale.<\/p><p>En effet, un mod\u00e8le sous-ajust\u00e9 s\u2019adapte mal au training set. Il pr\u00e9sentera des pr\u00e9dictions biais\u00e9es faute d\u2019entra\u00eenement suffisant et laissera appara\u00eetre de nombreuses erreurs en phase d\u2019apprentissage.<\/p><p>En d\u2019autres termes, c\u2019est un mod\u00e8le d\u2019apprentissage trop g\u00e9n\u00e9raliste qui ne parvient pas \u00e0 fournir des <a href=\"https:\/\/datascientest.com\/quest-ce-que-cest-le-predictive-modelling\">analyses pr\u00e9dictives<\/a> pr\u00e9cises.<\/p><p>Il est donc primordial de ma\u00eetriser ce concept en Machine Learning avant de concevoir des <strong>mod\u00e8les ad\u00e9quats<\/strong>. Mais avant de voir comment \u00e9viter l\u2019underfitting, nous allons aussi \u00e9tudier son contraire.<\/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-53a1515 e-flex e-con-boxed e-con\" data-id=\"53a1515\" 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-b86dd1e elementor-widget elementor-widget-heading\" data-id=\"b86dd1e\" 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\">L\u2019overfitting, antinomique de l\u2019underfitting<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e1f4c0d elementor-widget elementor-widget-text-editor\" data-id=\"e1f4c0d\" 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>\u00c0 l\u2019inverse de l&rsquo;underfitting, l\u2019overfitting se traduit par le <strong>sur-ajustement ou sur-apprentissage<\/strong>.<\/p><p>Ici, on utilise un <strong>maximum de variables pour \u00eatre pr\u00e9par\u00e9 \u00e0 toutes les hypoth\u00e8ses<\/strong>. Mais attention, le risque est d\u2019utiliser trop de variables non pertinentes par rapport \u00e0 l\u2019\u00e9tude en cours.<\/p><p>Si cette m\u00e9thode permet d\u2019obtenir des niveaux d\u2019erreur en apprentissage et en validation similaire, cela complexifie fortement le mod\u00e8le. Ce qui implique une augmentation du nombre de param\u00e8tres \u00e0 estimer. De ce fait, le mod\u00e8le va d\u00e9tecter plus de corr\u00e9lations fallacieuses.<\/p><p>Le mod\u00e8le sur-ajust\u00e9 s\u2019attache \u00e0 la m\u00e9morisation des particularit\u00e9s tr\u00e8s sp\u00e9cifiques des donn\u00e9es. On dit qu\u2019<strong>il s\u2019ajuste au bruit dans les donn\u00e9es<\/strong>. \u00c0 terme, il trouvera des liens de causalit\u00e9 entre des donn\u00e9es qui n\u2019ont aucun rapport. Il pourra alors se montrer tr\u00e8s performant sur les donn\u00e9es d\u2019apprentissage, mais cela au d\u00e9triment de la tendance g\u00e9n\u00e9rale.<\/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-8a7058e elementor-widget elementor-widget-image\" data-id=\"8a7058e\" 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=\"1055\" height=\"629\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/underfitting-vs-overfitting-e1670513276723.png\" class=\"attachment-full size-full wp-image-184128\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/underfitting-vs-overfitting-e1670513276723.png 1055w, https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/underfitting-vs-overfitting-e1670513276723-300x179.png 300w, https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/underfitting-vs-overfitting-e1670513276723-1024x611.png 1024w, https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/underfitting-vs-overfitting-e1670513276723-768x458.png 768w\" sizes=\"(max-width: 1055px) 100vw, 1055px\" \/>\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\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-022300d e-flex e-con-boxed e-con\" data-id=\"022300d\" 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-55b69f0 elementor-widget elementor-widget-heading\" data-id=\"55b69f0\" 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 \u00e9viter l\u2019underfitting ?<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ce43546 elementor-widget elementor-widget-text-editor\" data-id=\"ce43546\" 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>L\u2019enjeu pour les <a href=\"https:\/\/datascientest.com\/les-outils-dans-le-quotidien-dun-data-scientist\">Data Scientists<\/a> est de concevoir le mod\u00e8le ad\u00e9quat, ni trop simple, ni trop complexe. \u00c0 savoir, \u00e9viter l\u2019underfitting, sans tomber dans l\u2019overfitting.<\/p><p>Pour cela, il convient d\u2019identifier le sweet spot permettant de trouver le juste \u00e9quilibre.\u00a0Mais avant cela, vous devez apprendre \u00e0 <strong>d\u00e9tecter l\u2019underfitting et l\u2019overfitting<\/strong>.<\/p><p>Si les donn\u00e9es de test ont un faible taux d\u2019erreur, mais que les donn\u00e9es d&rsquo;entra\u00eenement ont un taux d\u2019erreur \u00e9lev\u00e9, il y a probablement un sous-ajustement.<\/p><p>Dans ce cas, attention \u00e0 ne pas trop complexifier le mod\u00e8le. Ici, c\u2019est la situation inverse qui se produira, \u00e0 savoir : des<strong> donn\u00e9es d&rsquo;entra\u00eenement<\/strong> avec un faible taux d\u2019erreur et des <strong>donn\u00e9es de tests<\/strong> avec un taux d\u2019erreur \u00e9lev\u00e9. L\u2019overfitting est caract\u00e9ris\u00e9.<\/p><p>D\u00e8s lors que vous \u00eates en mesure d\u2019identifier l\u2019un ou l\u2019autre, vous pouvez <strong>affiner le mod\u00e8le pr\u00e9dictif<\/strong> lors de la phase d\u2019apprentissage. Cela permet de r\u00e9duire progressivement les erreurs sur le training set.<\/p><p>Les Data Scientists doivent continuer \u00e0 affiner le mod\u00e8le jusqu\u2019\u00e0 ce que les erreurs augmentent dans la phase de validation. L\u2019\u00e9quilibre se situe juste avant que ces erreurs n\u2019augmentent \u00e0 cette \u00e9tape.<\/p><p>En appliquant cette m\u00e9thode, vous pourrez <strong>d\u00e9tecter la tendance dominante<\/strong> et l\u2019appliquer largement aux nouveaux ensembles de donn\u00e9es.<\/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-45c8f27 e-flex e-con-boxed e-con\" data-id=\"45c8f27\" 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-8f67ed5 elementor-widget elementor-widget-heading\" data-id=\"8f67ed5\" 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\">DataScientest pour ma\u00eetriser les mod\u00e8les de Machine Learning<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-8523010 elementor-widget elementor-widget-text-editor\" data-id=\"8523010\" 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>Trouver le juste \u00e9quilibre entre l\u2019underfitting et l\u2019overfitting n\u00e9cessite de la pratique et des connaissances accrues en <a href=\"https:\/\/datascientest.com\/data-analysis-tout-savoir\">analyse des donn\u00e9es<\/a>.<\/p><p>C\u2019est justement pour cette raison que DataScientest propose diff\u00e9rentes formations en Machine Learning. N\u2019h\u00e9sitez pas \u00e0 <strong>consulter nos programmes<\/strong>.<\/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-0e8fd79 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"0e8fd79\" 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 de 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\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-007486c e-flex e-con-boxed e-con\" data-id=\"007486c\" 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-01eedec elementor-widget elementor-widget-html\" data-id=\"01eedec\" data-element_type=\"widget\" data-widget_type=\"html.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<script type=\"application\/ld+json\">\r\n{\r\n  \"@context\": \"https:\/\/schema.org\",\r\n  \"@type\": \"FAQPage\",\r\n  \"mainEntity\": [{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"Qu\u2019est-ce que l\u2019underfitting ?\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"L\u2019underfitting est une notion incontournable du Machine Learning, puisqu\u2019il peut \u00eatre responsable d\u2019une mauvaise performance de l\u2019apprentissage supervis\u00e9 \u00e0 partir des donn\u00e9es. Alors avant de voir plus en d\u00e9tail cette notion, il convient de rappeler quelques \u00e9l\u00e9ments essentiels du fonctionnement du Machine Learning.\"\r\n    }\r\n  },{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"Comment \u00e9viter l\u2019underfitting ?\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"L\u2019enjeu pour les Data Scientists est de concevoir le mod\u00e8le ad\u00e9quat, ni trop simple, ni trop complexe. \u00c0 savoir, \u00e9viter l\u2019underfitting, sans tomber dans l\u2019overfitting. Pour cela, il convient d\u2019identifier le sweet spot permettant de trouver le juste \u00e9quilibre. Mais avant cela, vous devez apprendre \u00e0 d\u00e9tecter l\u2019underfitting et l\u2019overfitting. Si les donn\u00e9es de test ont un faible taux d\u2019erreur, mais que les donn\u00e9es d\u2019entra\u00eenement ont un taux d\u2019erreur \u00e9lev\u00e9, il y a probablement un sous-ajustement.\"\r\n    }\r\n  }]\r\n}\r\n<\/script>\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>La d\u00e9finition du bon mod\u00e8le est primordiale pour r\u00e9aliser des pr\u00e9dictions pertinentes en Machine Learning. Mais un mauvais ajustement des donn\u00e9es d\u2019apprentissage peut affecter la performance des analyses pr\u00e9dictives. C\u2019est justement ce qui se passe avec l\u2019underfitting. Alors de quoi s\u2019agit-il ? Et comment l\u2019\u00e9viter ? Les r\u00e9ponses sont ici. Qu\u2019est-ce que l\u2019underfitting ? L\u2019underfitting [&hellip;]<\/p>\n","protected":false},"author":77,"featured_media":184135,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"content-type":"","inline_featured_image":false},"categories":[2497],"tags":[],"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>Underfitting ou le sous-ajustement en Machine Learning<\/title>\n<meta name=\"description\" content=\"L\u2019underfitting est un mauvais ajustement des donn\u00e9es d\u2019apprentissage qui affecte la performance des analyses pr\u00e9dictives. Les d\u00e9tails dans cet article.\" \/>\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\/underfitting-tout-savoir\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Underfitting ou le sous-ajustement en Machine Learning\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/underfitting-tout-savoir\" \/>\n<meta property=\"og:site_name\" content=\"Formation Data Science | DataScientest.com\" \/>\n<meta property=\"article:published_time\" content=\"2022-12-20T16:32:46+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-11-09T13:55:10+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/couv-underfitting.png\" \/>\n\t<meta property=\"og:image:width\" content=\"2048\" \/>\n\t<meta property=\"og:image:height\" content=\"1125\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Raphael Kassel\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Raphael Kassel\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"5 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/datascientest.com\/underfitting-tout-savoir\",\"url\":\"https:\/\/datascientest.com\/underfitting-tout-savoir\",\"name\":\"Underfitting ou le sous-ajustement en Machine Learning\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/#website\"},\"datePublished\":\"2022-12-20T16:32:46+00:00\",\"dateModified\":\"2023-11-09T13:55:10+00:00\",\"author\":{\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\"},\"description\":\"L\u2019underfitting est un mauvais ajustement des donn\u00e9es d\u2019apprentissage qui affecte la performance des analyses pr\u00e9dictives. Les d\u00e9tails dans cet article.\",\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/underfitting-tout-savoir\"]}]},{\"@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\/6eca16f63daaafcbc96e15549b9d4ee1\",\"name\":\"Raphael Kassel\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"contentUrl\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"caption\":\"Raphael Kassel\"},\"description\":\"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \\\"Chef de projet IA\\\" et \\\"Data Product Manager\\\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/raphaelkassel\/\",\"Data Science\"],\"url\":\"https:\/\/datascientest.com\/author\/raphaelkassel\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"Underfitting ou le sous-ajustement en Machine Learning","description":"L\u2019underfitting est un mauvais ajustement des donn\u00e9es d\u2019apprentissage qui affecte la performance des analyses pr\u00e9dictives. Les d\u00e9tails dans cet article.","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\/underfitting-tout-savoir","og_locale":"fr_FR","og_type":"article","og_title":"Underfitting ou le sous-ajustement en Machine Learning","og_url":"https:\/\/datascientest.com\/underfitting-tout-savoir","og_site_name":"Formation Data Science | DataScientest.com","article_published_time":"2022-12-20T16:32:46+00:00","article_modified_time":"2023-11-09T13:55:10+00:00","og_image":[{"width":2048,"height":1125,"url":"https:\/\/datascientest.com\/wp-content\/uploads\/2022\/12\/couv-underfitting.png","type":"image\/png"}],"author":"Raphael Kassel","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Raphael Kassel","Dur\u00e9e de lecture estim\u00e9e":"5 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/datascientest.com\/underfitting-tout-savoir","url":"https:\/\/datascientest.com\/underfitting-tout-savoir","name":"Underfitting ou le sous-ajustement en Machine Learning","isPartOf":{"@id":"https:\/\/datascientest.com\/#website"},"datePublished":"2022-12-20T16:32:46+00:00","dateModified":"2023-11-09T13:55:10+00:00","author":{"@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1"},"description":"L\u2019underfitting est un mauvais ajustement des donn\u00e9es d\u2019apprentissage qui affecte la performance des analyses pr\u00e9dictives. Les d\u00e9tails dans cet article.","inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/underfitting-tout-savoir"]}]},{"@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\/6eca16f63daaafcbc96e15549b9d4ee1","name":"Raphael Kassel","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/datascientest.com\/#\/schema\/person\/image\/","url":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","contentUrl":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","caption":"Raphael Kassel"},"description":"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \"Chef de projet IA\" et \"Data Product Manager\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.","sameAs":["https:\/\/www.linkedin.com\/in\/raphaelkassel\/","Data Science"],"url":"https:\/\/datascientest.com\/author\/raphaelkassel"}]}},"_links":{"self":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/184112"}],"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\/77"}],"replies":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/comments?post=184112"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/184112\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media\/184135"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media?parent=184112"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/categories?post=184112"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/tags?post=184112"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}