{"id":20942,"date":"2024-02-21T07:32:43","date_gmt":"2024-02-21T06:32:43","guid":{"rendered":"https:\/\/complex-systems-ai.com\/?page_id=20942"},"modified":"2024-02-21T20:27:26","modified_gmt":"2024-02-21T19:27:26","slug":"forecasting-autogluon","status":"publish","type":"page","link":"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/","title":{"rendered":"Previsi\u00f3n con AutoGluon Amazon"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-page\" data-elementor-id=\"20942\" class=\"elementor elementor-20942\">\n\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d824177 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d824177\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-33 elementor-top-column elementor-element elementor-element-08322c1\" data-id=\"08322c1\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-470e13b elementor-align-justify elementor-widget elementor-widget-button\" data-id=\"470e13b\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm\" href=\"https:\/\/complex-systems-ai.com\/prediction-forecasting\/\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t\t\t\t<span class=\"elementor-button-text\">Forecasting<\/span>\n\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/div>\n\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<\/div>\n\t\t\t\t<div class=\"elementor-column elementor-col-33 elementor-top-column elementor-element elementor-element-adcb25e\" data-id=\"adcb25e\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-06c3522 elementor-align-justify elementor-widget elementor-widget-button\" data-id=\"06c3522\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm\" href=\"https:\/\/complex-systems-ai.com\/\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t\t\t\t<span class=\"elementor-button-text\">Page d'accueil<\/span>\n\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/div>\n\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<\/div>\n\t\t\t\t<div class=\"elementor-column elementor-col-33 elementor-top-column elementor-element elementor-element-806c046\" data-id=\"806c046\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a08ecaa elementor-align-justify elementor-widget elementor-widget-button\" data-id=\"a08ecaa\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm\" href=\"https:\/\/plat.ai\/blog\/difference-between-prediction-and-forecast\/\" target=\"_blank\" rel=\"noopener\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t\t\t\t<span class=\"elementor-button-text\">Wiki<\/span>\n\t\t\t\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t\t\t<\/div>\n\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-a5f9034 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"a5f9034\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-1cb27a3\" data-id=\"1cb27a3\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a19b8c0 elementor-widget elementor-widget-heading\" data-id=\"a19b8c0\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_82_2 counter-hierarchy ez-toc-counter ez-toc-grey ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">Contenus<\/p>\n<span class=\"ez-toc-title-toggle\"><a href=\"#\" class=\"ez-toc-pull-right ez-toc-btn ez-toc-btn-xs ez-toc-btn-default ez-toc-toggle\" aria-label=\"Alternar tabla de contenidos\"><span class=\"ez-toc-js-icon-con\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/span><\/a><\/span><\/div>\n<nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Forecasting-de-serie-temporelle-avec-AutoGluon\" >Forecasting de s\u00e9rie temporelle avec AutoGluon<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Introduction-a-AutoGluon\" >Introduction \u00e0 AutoGluon<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Comment-entrainer-le-predicateur\" >Comment entrainer le pr\u00e9dicateur<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Voir-les-resultats\" >Voir les r\u00e9sultats<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Utilisation-de-DeepAR\" >Utilisation de DeepAR<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Liste-des-methodes\" >Liste des m\u00e9thodes<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/complex-systems-ai.com\/es\/pronostico-de-prediccion\/pronostico-de-autogluones\/#Choix-des-metriques\" >Choix des m\u00e9triques<\/a><\/li><\/ul><\/nav><\/div>\n<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Forecasting-de-serie-temporelle-avec-AutoGluon\"><\/span>Forecasting de s\u00e9rie temporelle avec AutoGluon<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-b5f8e47 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"b5f8e47\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-d496909\" data-id=\"d496909\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-e493cc5 elementor-widget elementor-widget-text-editor\" data-id=\"e493cc5\" data-element_type=\"widget\" data-e-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\t\t<p>Introduction \u00e0 la biblioth\u00e8que Multimodale AutoML d&rsquo;Amazon AutoGluon avec un probl\u00e8me de pr\u00e9vision de s\u00e9ries chronologiques.<\/p><p><img decoding=\"async\" class=\"aligncenter wp-image-11096 size-full\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2020\/09\/cropped-Capture.png\" alt=\"AutoGluon\" width=\"97\" height=\"97\" title=\"\"><\/p>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-84a1e05 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"84a1e05\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-e8dd00a\" data-id=\"e8dd00a\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-dee8a68 elementor-widget elementor-widget-heading\" data-id=\"dee8a68\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Introduction-a-AutoGluon\"><\/span>Introduction \u00e0 AutoGluon<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-96a2d7d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"96a2d7d\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-dcd4b7d\" data-id=\"dcd4b7d\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a3041cf elementor-widget elementor-widget-text-editor\" data-id=\"a3041cf\" data-element_type=\"widget\" data-e-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\t\t<p><img fetchpriority=\"high\" decoding=\"async\" class=\"alignnone wp-image-20947 size-full\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon.webp\" alt=\"AutoGluon\" width=\"720\" height=\"440\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon.webp 720w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon-300x183.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon-18x12.webp 18w\" sizes=\"(max-width: 720px) 100vw, 720px\" \/><\/p><p>AutoGluon est une biblioth\u00e8que python multimodale open source pour AutoML, lanc\u00e9e par Amazon. Construite sur PyTorch, la biblioth\u00e8que utilise des mod\u00e8les de pointe (SOTA) pour obtenir les mod\u00e8les les plus performants pour divers probl\u00e8mes d&rsquo;apprentissage automatique.<\/p><p>\u00c9quip\u00e9 de mod\u00e8les SOTA Deep Learning, AutoGluon propose des solutions \u00e0 des probl\u00e8mes tels que la classification d&rsquo;images, la d\u00e9tection d&rsquo;objets, la pr\u00e9diction de texte, la segmentation d&rsquo;images, la pr\u00e9vision de s\u00e9ries chronologiques et bien plus encore, comme dans l&rsquo;image de couverture. Cette page vise \u00e0 montrer aux lecteurs comment utiliser AutoGluon pour une t\u00e2che de pr\u00e9vision de s\u00e9ries chronologiques en consid\u00e9rant le c\u00e9l\u00e8bre ensemble de donn\u00e9es Airlines avec un minimum de code.<\/p><p>Il est conseill\u00e9 de cr\u00e9er un nouvel environnement Python pour les exp\u00e9riences Autogluon afin d&rsquo;\u00e9viter les conflits avec d&rsquo;autres biblioth\u00e8ques. Une fois le nouvel environnement cr\u00e9\u00e9, installez Autogluon en ex\u00e9cutant la commande ci-dessous.<\/p><p>pip install autogluon<br \/>import autogluon<\/p><p>AutoGluon TimeSeriesPredictor attend des donn\u00e9es dans un format sp\u00e9cifique avec multi-index. Un index doit \u00eatre l&rsquo;horodatage et l&rsquo;autre doit \u00eatre un identifiant unique (il peut s&rsquo;agir de n&rsquo;importe quelle valeur). Une trame de donn\u00e9es pandas normale peut \u00eatre convertie dans ce format \u00e0 l&rsquo;aide de TimeSeriesDataFrame.from_data_frame().<\/p><p><span class=\"hljs-keyword\">from<\/span> autogluon.timeseries <span class=\"hljs-keyword\">import<\/span> TimeSeriesPredictor, TimeSeriesDataFrame<\/p><p>Commen\u00e7ons par charger les donn\u00e9es et les placer dans un DataFrame de s\u00e9rie temporelle :<\/p><p><span class=\"hljs-keyword\">import<\/span> pandas <span class=\"hljs-keyword\">as<\/span> pd<br \/><br \/>data = pd.read_excel(<span class=\"hljs-string\">&lsquo;airlinesData.xlsx&rsquo;<\/span>)<br \/>data[<span class=\"hljs-string\">&lsquo;Month&rsquo;<\/span>] = pd.to_datetime(data[<span class=\"hljs-string\">&lsquo;Month&rsquo;<\/span>]) <span class=\"hljs-comment\">#convert to datetime<\/span><br \/><span class=\"hljs-built_in\">id<\/span> = [<span class=\"hljs-string\">&lsquo;airline&rsquo;<\/span>] * <span class=\"hljs-built_in\">len<\/span>(data) <span class=\"hljs-comment\">#create list of id to add to dataframe<\/span><br \/>data[<span class=\"hljs-string\">&lsquo;id&rsquo;<\/span>] = <span class=\"hljs-built_in\">id<\/span> <span class=\"hljs-comment\">#add new colum &lsquo;id&rsquo; to dataframe<\/span><\/p><p>data = TimeSeriesDataFrame.from_data_frame(<br \/>data,<br \/>id_column=\u00a0\u00bbid\u00a0\u00bb,<br \/>timestamp_column=\u00a0\u00bbMonth\u00a0\u00bb<br \/>)<br \/>data.tail()<\/p><p>Puis on split les donn\u00e9es (attention ici on n&rsquo;a pas fait de cross-validation) :<\/p><p>data.shape<br \/>(96, 3)<br \/>data.columns<br \/>Index([&lsquo;Passengers&rsquo;], dtype=&rsquo;object&rsquo;)<br \/>\u00a0\u00bb&rsquo;split data into train and test\u00a0\u00bb&rsquo;<br \/>train = data.head(77)<br \/>test = data.tail(19)<\/p>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-c58e8aa elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c58e8aa\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-3ac5b04\" data-id=\"3ac5b04\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-2b8d2cf elementor-widget elementor-widget-heading\" data-id=\"2b8d2cf\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Comment-entrainer-le-predicateur\"><\/span>Comment entrainer le pr\u00e9dicateur<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-e4a4773 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e4a4773\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-3014ed0\" data-id=\"3014ed0\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-4376321 elementor-widget elementor-widget-text-editor\" data-id=\"4376321\" data-element_type=\"widget\" data-e-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\t\t<p>La m\u00e9trique d&rsquo;\u00e9valuation choisie d\u00e9pend si vous avez besoin d&rsquo;une pr\u00e9vision probabiliste ou d&rsquo;une pr\u00e9vision ponctuelle. Un tableau contenant diff\u00e9rentes mesures est pr\u00e9sent\u00e9 ci-dessous.<\/p><p><img decoding=\"async\" class=\"alignnone wp-image-20949 size-full\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon1.webp\" alt=\"autogluon\" width=\"952\" height=\"486\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon1.webp 952w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon1-300x153.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon1-768x392.webp 768w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon1-18x9.webp 18w\" sizes=\"(max-width: 952px) 100vw, 952px\" \/><\/p><p>Diff\u00e9rents mod\u00e8les form\u00e9s sur les donn\u00e9es du train, les param\u00e8tres choisis et la m\u00e9trique d&rsquo;\u00e9valuation peuvent \u00eatre consult\u00e9s dans le journal de formation. La biblioth\u00e8que exploite des mod\u00e8les statistiques et des mod\u00e8les SOTA Deep Learning pour la formation. Par d\u00e9faut, un mod\u00e8le WeightedEnsemble sera \u00e9galement essay\u00e9 par le mod\u00e8le. il peut \u00eatre d\u00e9sactiv\u00e9 en d\u00e9finissant activate_ensemble = False dans la m\u00e9thode .fit().<\/p><pre class=\"mc md me mf mg ot os ou bo ov ba bj\"><span id=\"4c9f\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">predictor = TimeSeriesPredictor(target='Passengers', <br \/>                                prediction_length=19,<br \/>                                 eval_metric=\"MASE\",).fit(train)<\/span><\/pre><pre class=\"pd ot os ou bo ov ba bj\"><span id=\"053f\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">No path specified. Models will be saved in: \"AutogluonModels\\ag-20240202_034220\"<br \/>Beginning AutoGluon training...<br \/>AutoGluon will save models to 'AutogluonModels\\ag-20240202_034220'<br \/>=================== System Info ===================<br \/>AutoGluon Version:  1.0.0<br \/>Python Version:     3.8.18<br \/>Operating System:   Windows<br \/>Platform Machine:   AMD64<br \/>Platform Version:   10.0.22621<br \/>CPU Count:          12<br \/>GPU Count:          0<br \/>Memory Avail:       0.59 GB \/ 7.33 GB (8.0%)<br \/>Disk Space Avail:   262.60 GB \/ 476.08 GB (55.2%)<br \/>===================================================<br \/><br \/>Fitting with arguments:<br \/>{'enable_ensemble': True,<br \/> 'eval_metric': MASE,<br \/> 'hyperparameters': 'default',<br \/> 'known_covariates_names': [],<br \/> 'num_val_windows': 1,<br \/> 'prediction_length': 19,<br \/> 'quantile_levels': [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9],<br \/> 'random_seed': 123,<br \/> 'refit_every_n_windows': 1,<br \/> 'refit_full': False,<br \/> 'target': 'Passengers',<br \/> 'verbosity': 2}<br \/><br \/>Inferred time series frequency: 'MS'<br \/>Provided train_data has 77 rows, 1 time series. Median time series length is 77 (min=77, max=77). <br \/><br \/>Provided dataset contains following columns:<br \/>\ttarget:           'Passengers'<br \/><br \/>AutoGluon will gauge predictive performance using evaluation metric: 'MASE'<br \/>\tThis metric's sign has been flipped to adhere to being higher_is_better. The metric score can be multiplied by -1 to get the metric value.<br \/>===================================================<br \/><br \/>Starting training. Start time is 2024-02-02 09:12:20<br \/>Models that will be trained: ['SeasonalNaive', 'CrostonSBA', 'NPTS', 'AutoETS', 'DynamicOptimizedTheta', 'AutoARIMA', 'RecursiveTabular', 'DirectTabular', 'DeepAR', 'TemporalFusionTransformer', 'PatchTST']<br \/>Training timeseries model SeasonalNaive. <br \/>\t-0.8728       = Validation score (-MASE)<br \/>\t0.02    s     = Training runtime<br \/>\t6.75    s     = Validation (prediction) runtime<br \/>Training timeseries model CrostonSBA. <br \/>\t-1.4745       = Validation score (-MASE)<br \/>\t0.00    s     = Training runtime<br \/>\t15.24   s     = Validation (prediction) runtime<br \/>Training timeseries model NPTS. <br \/>\t-2.3546       = Validation score (-MASE)<br \/>\t0.02    s     = Training runtime<br \/>\t2.85    s     = Validation (prediction) runtime<br \/>Training timeseries model AutoETS. <br \/>\t-0.8152       = Validation score (-MASE)<br \/>\t0.02    s     = Training runtime<br \/>\t30.67   s     = Validation (prediction) runtime<br \/>Training timeseries model DynamicOptimizedTheta. <br \/>\t-1.4460       = Validation score (-MASE)<br \/>\t0.02    s     = Training runtime<br \/>\t29.97   s     = Validation (prediction) runtime<br \/>Training timeseries model AutoARIMA. <br \/>\t-0.7476       = Validation score (-MASE)<br \/>\t0.02    s     = Training runtime<br \/>\t21.59   s     = Validation (prediction) runtime<br \/>Training timeseries model RecursiveTabular. <br \/>\t-0.5560       = Validation score (-MASE)<br \/>\t4.89    s     = Training runtime<br \/>\t0.38    s     = Validation (prediction) runtime<br \/>Training timeseries model DirectTabular. <br \/>\t-3.5581       = Validation score (-MASE)<br \/>\t0.65    s     = Training runtime<br \/>\t0.08    s     = Validation (prediction) runtime<br \/>Training timeseries model DeepAR. <br \/>\t-1.1489       = Validation score (-MASE)<br \/>\t55.81   s     = Training runtime<br \/>\t0.08    s     = Validation (prediction) runtime<br \/>Training timeseries model TemporalFusionTransformer. <br \/>\t-0.8554       = Validation score (-MASE)<br \/>\t144.14  s     = Training runtime<br \/>\t0.02    s     = Validation (prediction) runtime<br \/>Training timeseries model PatchTST. <br \/>\t-0.9520       = Validation score (-MASE)<br \/>\t32.19   s     = Training runtime<br \/>\t0.03    s     = Validation (prediction) runtime<br \/>Fitting simple weighted ensemble.<br \/>\tEnsemble weights: {'AutoARIMA': 0.09, 'NPTS': 0.07, 'RecursiveTabular': 0.04, 'SeasonalNaive': 0.35, 'TemporalFusionTransformer': 0.44}<br \/>\t-0.2127       = Validation score (-MASE)<br \/>\t1.67    s     = Training runtime<br \/>\t31.59   s     = Validation (prediction) runtime<br \/>Training complete. Models trained: ['SeasonalNaive', 'CrostonSBA', 'NPTS', 'AutoETS', 'DynamicOptimizedTheta', 'AutoARIMA', 'RecursiveTabular', 'DirectTabular', 'DeepAR', 'TemporalFusionTransformer', 'PatchTST', 'WeightedEnsemble']<br \/>Total runtime: 347.54 s<br \/>Best model: WeightedEnsemble<br \/>Best model score: -0.2127<\/span><\/pre>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-2cf2886 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2cf2886\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-d270fc2\" data-id=\"d270fc2\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-b1a4710 elementor-widget elementor-widget-heading\" data-id=\"b1a4710\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Voir-les-resultats\"><\/span>Voir les r\u00e9sultats<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-3a56c61 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"3a56c61\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-7578d19\" data-id=\"7578d19\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-6651bd3 elementor-widget elementor-widget-text-editor\" data-id=\"6651bd3\" data-element_type=\"widget\" data-e-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\t\t<p>Pour cela, il suffit d&rsquo;utiliser la fonction :<\/p><p>predictor.leaderboard()<\/p><p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-20950 size-large\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2-1024x502.webp\" alt=\"autogluon\" width=\"1024\" height=\"502\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2-1024x502.webp 1024w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2-300x147.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2-768x376.webp 768w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2-1536x753.webp 1536w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2-18x9.webp 18w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon2.webp 1620w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/p><p>Sans sp\u00e9cifier de mod\u00e8le, le pr\u00e9dicteur consid\u00e9rera le meilleur mod\u00e8le de pr\u00e9diction qui est WeightedEnsemble dans ce cas.<\/p><pre class=\"mc md me mf mg ot os ou bo ov ba bj\"><span id=\"15e9\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">predictions = predictor.predict(train)\npredictions <\/span><\/pre><pre class=\"pd ot os ou bo ov ba bj\"><span id=\"244f\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">Model not specified in predict, will default to the model with the best validation score: WeightedEnsemble<\/span><\/pre><figure class=\"mc md me mf mg mh lz ma paragraph-image\"><div class=\"ab cm ca pb\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-20951 size-large\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3-1024x551.webp\" alt=\"autogluon\" width=\"1024\" height=\"551\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3-1024x551.webp 1024w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3-300x162.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3-768x413.webp 768w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3-1536x827.webp 1536w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3-18x10.webp 18w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon3.webp 1930w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/div><div><p>Par d\u00e9faut, AutoGluon pr\u00e9dit les niveaux quantiles [0,1, 0,2, 0,3, 0,4, 0,5, 0,6, 0,7, 0,8, 0,9]. Pour pr\u00e9dire un ensemble diff\u00e9rent de quantiles, vous pouvez utiliser les arguments quantile_levels comme\u00a0:<\/p><p>pr\u00e9dicteur = TimeSeriesPredictor(eval_metric=\u201dWQL\u201d,quantile_levels=[0.1, 0.5, 0.75, 0.9]) [2].<\/p><p>Il est aussi possible d&rsquo;afficher la pr\u00e9diction :<\/p><pre class=\"mc md me mf mg ot os ou bo ov ba bj\"><span id=\"6114\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">import matplotlib.pyplot as plt\n\nplt.figure(figsize=(20, 3))\n\nitem_id = \"airline\"\ny_past = train.loc[item_id][\"Passengers\"]\ny_pred = predictions.loc[item_id]\ny_test = test.loc[item_id][\"Passengers\"]\n\nplt.plot(y_past, label=\"Past time series values\")\nplt.plot(y_pred[\"mean\"], label=\"Mean forecast\")\nplt.plot(y_test, label=\"Future time series values\")\n\nplt.fill_between(\n    y_pred.index, y_pred[\"0.1\"], y_pred[\"0.9\"], color=\"red\", alpha=0.1, label=f\"10%-90% confidence interval\"\n)\nplt.legend();<\/span><\/pre><figure class=\"mc md me mf mg mh lz ma paragraph-image\"><div class=\"ab cm ca pb\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-20952 size-large\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4-1024x175.webp\" alt=\"autogluon\" width=\"1024\" height=\"175\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4-1024x175.webp 1024w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4-300x51.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4-768x131.webp 768w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4-1536x262.webp 1536w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4-18x3.webp 18w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon4.webp 1606w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/div><p>Pour utiliser un mod\u00e8le sp\u00e9cifique parmi les mod\u00e8les form\u00e9s, choisissez dans le classement ou les mod\u00e8les disponibles peuvent \u00eatre vus avec le morceau de code suivant.<\/p><\/figure><\/div><\/figure>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-0ea1586 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"0ea1586\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-f9f1dc3\" data-id=\"f9f1dc3\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-f6fc0ad elementor-widget elementor-widget-heading\" data-id=\"f6fc0ad\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Utilisation-de-DeepAR\"><\/span>Utilisation de DeepAR<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-5476e58 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"5476e58\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-1031584\" data-id=\"1031584\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a1dda44 elementor-widget elementor-widget-text-editor\" data-id=\"a1dda44\" data-element_type=\"widget\" data-e-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\t\t<p>Un mod\u00e8le pr\u00e9f\u00e9r\u00e9 peut \u00eatre choisi dans Model Zoo et utilis\u00e9. En choisissant DeepAR comme mod\u00e8le d&rsquo;int\u00e9r\u00eat, l&rsquo;entrainement peut \u00eatre effectu\u00e9e comme ci-dessous.<\/p><pre class=\"mc md me mf mg ot os ou bo ov ba bj\"><span id=\"7adf\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">predictor = TimeSeriesPredictor(target='Passengers', prediction_length=19).fit(<br \/>   train,<br \/>   hyperparameters={<br \/>      \"DeepAR\": {},<br \/>   },<br \/>)<\/span><\/pre><pre class=\"pd ot os ou bo ov ba bj\"><span class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">Beginning AutoGluon training...<br \/>AutoGluon will save models to 'AutogluonModels\\ag-20240201_193851'<br \/>=================== System Info ===================<br \/>AutoGluon Version:  1.0.0<br \/>Python Version:     3.8.18<br \/>Operating System:   Windows<br \/>Platform Machine:   AMD64<br \/>Platform Version:   10.0.22621<br \/>CPU Count:          12<br \/>GPU Count:          0<br \/>Memory Avail:       1.51 GB \/ 7.33 GB (20.6%)<br \/>Disk Space Avail:   262.48 GB \/ 476.08 GB (55.1%)<br \/>===================================================<\/span><\/pre><p>Fitting with arguments:<br \/>{&lsquo;enable_ensemble&rsquo;: True,<br \/>&lsquo;eval_metric&rsquo;: WQL,<br \/>&lsquo;hyperparameters&rsquo;: {&lsquo;DeepAR&rsquo;: {}},<br \/>&lsquo;known_covariates_names&rsquo;: [],<br \/>&lsquo;num_val_windows&rsquo;: 1,<br \/>&lsquo;prediction_length&rsquo;: 19,<br \/>&lsquo;quantile_levels&rsquo;: [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9],<br \/>&lsquo;random_seed&rsquo;: 123,<br \/>&lsquo;refit_every_n_windows&rsquo;: 1,<br \/>&lsquo;refit_full&rsquo;: False,<br \/>&lsquo;target&rsquo;: &lsquo;Passengers&rsquo;,<br \/>&lsquo;verbosity&rsquo;: 2}<\/p><p>Inferred time series frequency: &lsquo;MS&rsquo;<br \/>Provided train_data has 77 rows, 1 time series. Median time series length is 77 (min=77, max=77).<\/p><p>Provided dataset contains following columns:<br \/>target: &lsquo;Passengers&rsquo;<\/p><p>AutoGluon will gauge predictive performance using evaluation metric: &lsquo;WQL&rsquo;<br \/>This metric&rsquo;s sign has been flipped to adhere to being higher_is_better. The metric score can be multiplied by -1 to get the metric value.<br \/>===================================================<\/p><p>Starting training. Start time is 2024-02-02 01:08:51<br \/>Models that will be trained: [&lsquo;DeepAR&rsquo;]<br \/>Training timeseries model DeepAR. <br \/>-0.1003 = Validation score (-WQL)<br \/>50.19 s = Training runtime<br \/>0.11 s = Validation (prediction) runtime<br \/>Not fitting ensemble as only 1 model was trained.<br \/>Training complete. Models trained: [&lsquo;DeepAR&rsquo;]<br \/>Total runtime: 50.33 s<br \/>Best model: DeepAR<br \/>Best model score: -0.1003<\/p><pre class=\"pd ot os ou bo ov ba bj\"><span id=\"a1fb\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">predictions = predictor1.predict(train)<br \/>predictions<\/span><\/pre><pre class=\"pd ot os ou bo ov ba bj\"><span id=\"bfc1\" class=\"ow mu fr os b bf ox oy l oz pa\" data-selectable-paragraph=\"\">Model not specified in predict, will default to the model with the best validation score<\/span><\/pre><p>predictions = predictor.predict(train)<br \/>predictions<\/p><p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-20953 size-large\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5-1024x551.webp\" alt=\"autogluon\" width=\"1024\" height=\"551\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5-1024x551.webp 1024w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5-300x162.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5-768x413.webp 768w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5-1536x827.webp 1536w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5-18x10.webp 18w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon5.webp 1930w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/p><p>import matplotlib.pyplot as plt<\/p><p>plt.figure(figsize=(20, 3))<\/p><p>item_id = \u00ab\u00a0airline\u00a0\u00bb<br \/>y_past = train.loc[item_id][\u00ab\u00a0Passengers\u00a0\u00bb]<br \/>y_pred = predictions.loc[item_id]<br \/>y_test = test.loc[item_id][\u00ab\u00a0Passengers\u00a0\u00bb]<\/p><p>plt.plot(y_past, label=\u00a0\u00bbPast time series values\u00a0\u00bb)<br \/>plt.plot(y_pred[\u00ab\u00a0mean\u00a0\u00bb], label=\u00a0\u00bbMean forecast\u00a0\u00bb)<br \/>plt.plot(y_test, label=\u00a0\u00bbFuture time series values\u00a0\u00bb)<\/p><p>plt.fill_between(<br \/>y_pred.index, y_pred[\u00ab\u00a00.1\u00a0\u00bb], y_pred[\u00ab\u00a00.9&Prime;], color=\u00a0\u00bbred\u00a0\u00bb, alpha=0.1, label=f\u00a0\u00bb10%-90% confidence interval\u00a0\u00bb<br \/>)<br \/>plt.legend();<\/p><p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-20954 size-large\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6-1024x175.webp\" alt=\"autogluon\" width=\"1024\" height=\"175\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6-1024x175.webp 1024w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6-300x51.webp 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6-768x131.webp 768w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6-1536x262.webp 1536w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6-18x3.webp 18w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon6.webp 1606w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/p><p>Les mod\u00e8les s&rsquo;auto-tune pour trouver le meilleur param\u00e8trage. AutoGluon est un outil d&rsquo;autoML, l&rsquo;utilisateur n&rsquo;a pas besoin de tuner lui-m\u00eame les mod\u00e8les.<\/p>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-566e16c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"566e16c\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-2cf71bf\" data-id=\"2cf71bf\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-79bb1e1 elementor-widget elementor-widget-heading\" data-id=\"79bb1e1\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Liste-des-methodes\"><\/span>Liste des m\u00e9thodes<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-1de3e07 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"1de3e07\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-22a370c\" data-id=\"22a370c\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-dc4c37d elementor-widget elementor-widget-text-editor\" data-id=\"dc4c37d\" data-element_type=\"widget\" data-e-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\t\t<p>Mod\u00e8le na\u00eff &#8211; Mod\u00e8le de r\u00e9f\u00e9rence qui d\u00e9finit la pr\u00e9vision \u00e9gale \u00e0 la derni\u00e8re valeur observ\u00e9e.<\/p><p>SaisonnierNa\u00effMod\u00e8le &#8211; Mod\u00e8le de r\u00e9f\u00e9rence qui d\u00e9finit la pr\u00e9vision \u00e9gale \u00e0 la derni\u00e8re valeur observ\u00e9e pour la m\u00eame saison.<\/p><p>Mod\u00e8le moyen &#8211; Mod\u00e8le de r\u00e9f\u00e9rence qui d\u00e9finit la pr\u00e9vision \u00e9gale \u00e0 la moyenne ou au quantile historique.<\/p><p>Mod\u00e8leMoyenSaisonnier &#8211; Mod\u00e8le de r\u00e9f\u00e9rence qui d\u00e9finit la pr\u00e9vision \u00e9gale \u00e0 la moyenne historique ou au quantile de la m\u00eame saison.<\/p><p>Mod\u00e8le Z\u00e9ro &#8211; Un pr\u00e9visionniste na\u00eff qui renvoie toujours 0 pr\u00e9vision sur l\u2019horizon de pr\u00e9vision, o\u00f9 les intervalles de pr\u00e9vision sont calcul\u00e9s \u00e0 l\u2019aide d\u2019une pr\u00e9diction conforme.<\/p><p>Mod\u00e8le ETS &#8211; Lissage exponentiel avec tendance et saisonnalit\u00e9.<\/p><p>Mod\u00e8le AutoARIMA &#8211; Mod\u00e8le ARIMA r\u00e9gl\u00e9 automatiquement.<\/p><p>Mod\u00e8le AutoETS &#8211; Lissage exponentiel automatiquement ajust\u00e9 avec tendance et saisonnalit\u00e9.<\/p><p>Mod\u00e8le AutoCES &#8211; Pr\u00e9vision avec un mod\u00e8le de lissage exponentiel complexe o\u00f9 la s\u00e9lection du mod\u00e8le est effectu\u00e9e \u00e0 l&rsquo;aide du crit\u00e8re d&rsquo;information d&rsquo;Akaike.<\/p><p>Th\u00eataMod\u00e8le &#8211; Mod\u00e8le de pr\u00e9vision th\u00eata [Assimakopoulos2000].<\/p><p>ADIDAMod\u00e8le &#8211; Mod\u00e8le de pr\u00e9vision de la demande intermittente utilisant l\u2019approche de demande intermittente globale-d\u00e9sagr\u00e9g\u00e9e [Nikolopoulos2011].<\/p><p>CrostonClassicMod\u00e8le &#8211;\u00a0 Mod\u00e8le de pr\u00e9vision de la demande intermittente utilisant le mod\u00e8le de Croston o\u00f9 le param\u00e8tre de lissage est fix\u00e9 \u00e0 0,1 [Croston1972].<\/p><p>CrostonMod\u00e8le optimis\u00e9 &#8211; Mod\u00e8le de pr\u00e9vision de la demande intermittente utilisant le mod\u00e8le de Croston o\u00f9 le param\u00e8tre de lissage est optimis\u00e9 [Croston1972].<\/p><p>CrostonSBAMod\u00e8le &#8211; Mod\u00e8le de pr\u00e9vision de la demande intermittente utilisant le mod\u00e8le de Croston avec l&rsquo;approche de correction du biais Syntetos-Boylan [SyntetosBoylan2001].<\/p><p>Mod\u00e8le IMAPA &#8211; Mod\u00e8le de pr\u00e9vision de la demande intermittente utilisant l&rsquo;algorithme de pr\u00e9vision d&rsquo;agr\u00e9gation multiple intermittente [Petropoulos2015].<\/p><p>Mod\u00e8le NPTS &#8211; Pr\u00e9visionniste de s\u00e9ries chronologiques non param\u00e9triques.<\/p><p>Mod\u00e8le DeepAR &#8211; Mod\u00e8le de pr\u00e9vision autor\u00e9gressif bas\u00e9 sur un r\u00e9seau neuronal r\u00e9current [Salinas2020].<\/p><p>DLin\u00e9aireMod\u00e8le &#8211; R\u00e9seau neuronal simple \u00e0 r\u00e9troaction qui soustrait la tendance avant la pr\u00e9vision [Zeng2023].<\/p><p>PatchTSTMod\u00e8le &#8211; Pr\u00e9visionniste bas\u00e9 sur un transformateur qui segmente chaque s\u00e9rie temporelle en correctifs [Nie2023].<\/p><p>SimpleFeedForwardModel &#8211; R\u00e9seau neuronal simple \u00e0 r\u00e9troaction qui pr\u00e9dit simultan\u00e9ment toutes les valeurs futures.<\/p><p>Mod\u00e8le de transformateur de fusion temporelle &#8211; Combine LSTM avec une couche de transformateur pour pr\u00e9dire les quantiles de toutes les valeurs cibles futures [Lim2021].<\/p><p>Mod\u00e8le WaveNet &#8211; Estimateur WaveNet qui utilise l&rsquo;architecture propos\u00e9e dans [Oord2016] avec des cibles quantifi\u00e9es.<\/p><p>Mod\u00e8leTabulaire Direct &#8211; Pr\u00e9disez simultan\u00e9ment toutes les valeurs futures des s\u00e9ries chronologiques \u00e0 l\u2019aide de TabularPredictor d\u2019AutoGluon-Tabular.<\/p><p>Mod\u00e8leTabulaire R\u00e9cursif &#8211; Pr\u00e9disez les valeurs futures des s\u00e9ries chronologiques une par une \u00e0 l&rsquo;aide de TabularPredictor d&rsquo;AutoGluon-Tabular.<\/p>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-ad568c7 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"ad568c7\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-36826da\" data-id=\"36826da\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-16a71aa elementor-widget elementor-widget-heading\" data-id=\"16a71aa\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<h2 class=\"elementor-heading-title elementor-size-default\"><span class=\"ez-toc-section\" id=\"Choix-des-metriques\"><\/span>Choix des m\u00e9triques<span class=\"ez-toc-section-end\"><\/span><\/h2>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-0930b30 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"0930b30\" data-element_type=\"section\" data-e-type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-2304f36\" data-id=\"2304f36\" data-element_type=\"column\" data-e-type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t<div class=\"elementor-element elementor-element-2977451 elementor-widget elementor-widget-text-editor\" data-id=\"2977451\" data-element_type=\"widget\" data-e-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\t\t<p>Choisir la bonne m\u00e9trique d&rsquo;\u00e9valuation est l&rsquo;un des choix les plus importants lors de l&rsquo;utilisation d&rsquo;un framework AutoML. Cette page r\u00e9pertorie les m\u00e9triques d&rsquo;\u00e9valuation des pr\u00e9visions disponibles dans AutoGluon, explique quand diff\u00e9rentes m\u00e9triques doivent \u00eatre utilis\u00e9es et d\u00e9crit comment d\u00e9finir des m\u00e9triques d&rsquo;\u00e9valuation personnalis\u00e9es.<\/p><p>Lorsque vous utilisez AutoGluon, vous pouvez sp\u00e9cifier la m\u00e9trique \u00e0 l&rsquo;aide de l&rsquo;argument eval_metric de TimeSeriesPredictor, par exemple\u00a0:<\/p><p>from autogluon.timeseries import TimeSeriesPredictor<\/p><p>predictor = TimeSeriesPredictor(eval_metric=\u00a0\u00bbMASE\u00a0\u00bb)<\/p><p>AutoGluon rapporte toujours toutes les mesures dans un format \u00ab le plus \u00e9lev\u00e9 est le meilleur \u00bb. A cet effet, certaines m\u00e9triques sont multipli\u00e9es par -1. Par exemple, si nous d\u00e9finissons eval_metric=\u00a0\u00bbMASE\u00a0\u00bb, le pr\u00e9dicteur rapportera en fait -MASE (c&rsquo;est-\u00e0-dire le score MASE multipli\u00e9 par -1). Cela signifie que le test_score sera compris entre 0 (pr\u00e9vision la plus pr\u00e9cise) et (pr\u00e9vision la moins pr\u00e9cise).<\/p><p>Actuellement, AutoGluon prend en charge les m\u00e9triques d&rsquo;\u00e9valuation suivantes\u00a0:<\/p><p>SQL\u00a0Perte quantile mise \u00e0 l\u2019\u00e9chelle.<\/p><p>WQL\u00a0Perte quantile pond\u00e9r\u00e9e.<\/p><p>MAE\u00a0Erreur absolue moyenne.<\/p><p>MAPE\u00a0Erreur moyenne absolue en pourcentage.<\/p><p>MASE\u00a0Erreur d&rsquo;\u00e9chelle absolue moyenne.<\/p><p>MSE\u00a0Erreur quadratique moyenne.<\/p><p>RMSE\u00a0Erreur quadratique moyenne.<\/p><p>RMSSE\u00a0Erreur d&rsquo;\u00e9chelle quadratique moyenne.<\/p><p>SMAPE\u00a0Erreur de pourcentage absolu moyenne sym\u00e9trique.<\/p><p>WAPE\u00a0Erreur en pourcentage absolu pond\u00e9r\u00e9.<\/p><p>Vous pouvez \u00e9galement d\u00e9finir une m\u00e9trique d\u2019\u00e9valuation des pr\u00e9visions personnalis\u00e9e.<\/p><p>Si vous ne savez pas quelle mesure d&rsquo;\u00e9valuation choisir, voici trois questions qui peuvent vous aider \u00e0 faire le bon choix pour votre cas d&rsquo;utilisation.<\/p><p>1. Etes-vous int\u00e9ress\u00e9 par une pr\u00e9vision ponctuelle ou une pr\u00e9vision probabiliste ?<\/p><p>Si votre objectif est de g\u00e9n\u00e9rer une pr\u00e9vision probabiliste pr\u00e9cise, vous devez utiliser des m\u00e9triques WQL ou SQL. Ces m\u00e9triques sont bas\u00e9es sur la perte quantile et mesurent l\u2019exactitude des pr\u00e9visions quantiles. Par d\u00e9faut, AutoGluon pr\u00e9dit les niveaux quantiles [0,1, 0,2, 0,3, 0,4, 0,5, 0,6, 0,7, 0,8, 0,9]. Pour pr\u00e9dire un ensemble diff\u00e9rent de quantiles, vous pouvez utiliser l&rsquo;argument quantile_levels\u00a0:<\/p><p>pr\u00e9dicteur = TimeSeriesPredictor(eval_metric=\u00a0\u00bbWQL\u00a0\u00bb, quantile_levels=[0.1, 0.5, 0.75, 0.9])<\/p><p>Toutes les autres mesures de pr\u00e9vision d\u00e9crites sur cette page sont des mesures de pr\u00e9vision ponctuelles. Notez que si vous s\u00e9lectionnez eval_metric pour une m\u00e9trique de pr\u00e9vision ponctuelle lors de la cr\u00e9ation de TimeSeriesPredictor, la pr\u00e9vision minimisant cette m\u00e9trique sera toujours fournie dans la colonne \u00ab moyenne \u00bb du bloc de donn\u00e9es de pr\u00e9dictions.<\/p><p>2. Vous souciez-vous davantage de pr\u00e9dire avec pr\u00e9cision les s\u00e9ries chronologiques avec de grandes valeurs\u00a0?<\/p><p>Si la r\u00e9ponse est \u00ab oui \u00bb (par exemple, s\u2019il est important de pr\u00e9dire plus pr\u00e9cis\u00e9ment les ventes de produits populaires), vous devez utiliser des mesures d\u00e9pendant de l\u2019\u00e9chelle telles que WQL, MAE, RMSE ou WAPE. Ces m\u00e9triques sont \u00e9galement bien adapt\u00e9es au traitement de s\u00e9ries temporelles clairsem\u00e9es (intermittentes) comportant de nombreux z\u00e9ros.<\/p><p>Si la r\u00e9ponse est \u00ab non \u00bb (vous vous souciez de la m\u00eame mani\u00e8re de toutes les s\u00e9ries chronologiques de l&rsquo;ensemble de donn\u00e9es), envisagez des m\u00e9triques mises \u00e0 l&rsquo;\u00e9chelle telles que SQL, MASE et RMSSE. Alternativement, les mesures bas\u00e9es sur un pourcentage MAPE et SMAPE peuvent \u00e9galement \u00eatre utilis\u00e9es pour \u00e9galiser l&rsquo;\u00e9chelle entre les s\u00e9ries chronologiques. Cependant, ces mesures bas\u00e9es sur des pourcentages pr\u00e9sentent certaines limites bien document\u00e9es, nous ne recommandons donc pas de les utiliser dans la pratique. Notez que les mesures mises \u00e0 l&rsquo;\u00e9chelle et bas\u00e9es sur un pourcentage sont mal adapt\u00e9es aux donn\u00e9es clairsem\u00e9es (intermittentes).<\/p><p>3. (Pr\u00e9vision ponctuelle uniquement) Voulez-vous estimer la moyenne ou la m\u00e9diane\u00a0?<\/p><p>Pour estimer la m\u00e9diane, vous devez utiliser des m\u00e9triques telles que MAE, MASE ou WAPE. Si votre objectif est de pr\u00e9dire la moyenne (valeur attendue), vous devez utiliser les m\u00e9triques MSE, RMSE ou RMSSE.<\/p><p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone wp-image-20959 size-full\" src=\"http:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon7.png\" alt=\"autogluon\" width=\"674\" height=\"428\" title=\"\" srcset=\"https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon7.png 674w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon7-300x191.png 300w, https:\/\/complex-systems-ai.com\/wp-content\/uploads\/2024\/02\/autogluon7-18x12.png 18w\" sizes=\"(max-width: 674px) 100vw, 674px\" \/><\/p>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Pron\u00f3stico Inicio Wiki Pron\u00f3stico de series temporales con AutoGluon Introducci\u00f3n a la biblioteca multimodal AutoML de Amazon AutoGluon con un problema de pron\u00f3stico de series temporales... <\/p>","protected":false},"author":1,"featured_media":0,"parent":20753,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-20942","page","type-page","status-publish","hentry"],"amp_enabled":true,"_links":{"self":[{"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/pages\/20942","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/comments?post=20942"}],"version-history":[{"count":7,"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/pages\/20942\/revisions"}],"predecessor-version":[{"id":20962,"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/pages\/20942\/revisions\/20962"}],"up":[{"embeddable":true,"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/pages\/20753"}],"wp:attachment":[{"href":"https:\/\/complex-systems-ai.com\/es\/wp-json\/wp\/v2\/media?parent=20942"}],"curies":[{"name":"gracias","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}