{"id":1923,"date":"2019-09-02T05:42:23","date_gmt":"2019-09-02T05:42:23","guid":{"rendered":"http:\/\/kusuaks7\/?p=1528"},"modified":"2024-04-23T15:54:08","modified_gmt":"2024-04-23T15:54:08","slug":"a-simple-hands-on-tutorial-of-azure-machine-learning-studio","status":"publish","type":"post","link":"https:\/\/www.experfy.com\/blog\/ai-ml\/a-simple-hands-on-tutorial-of-azure-machine-learning-studio\/","title":{"rendered":"A simple hands-on tutorial of Azure Machine Learning Studio"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"1923\" class=\"elementor elementor-1923\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t<section class=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-26cd67cf elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"26cd67cf\" 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=\"has_eae_slider elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-6291235c\" data-id=\"6291235c\" 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-5c33aaa7 elementor-widget elementor-widget-text-editor\" data-id=\"5c33aaa7\" 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<section>\n<p id=\"6394\" data-selectable-paragraph=\"\">I\u2019ve recently stumbled upon a Microsoft Azure tool called\u00a0<strong>Microsoft Azure Machine Learning Studio<\/strong>, which is a graphical, web interface to perform machine learning operations using a\u00a0<strong>visual workflow<\/strong>, without the need of writing any code.<\/p>\n<p id=\"8a3d\" data-selectable-paragraph=\"\">I\u2019ve always been a\u00a0<strong>coder\u00a0<\/strong>and R has been my professional partner since the University period, so I\u2019ve always had little confidence in graphical software. When I discovered ML Studio and performed in a\u00a0<strong>few hours<\/strong>\u00a0what could reasonably take a few days of coding in R, I\u2019ve definitely added ML Studio to my Data Science Toolbox.<\/p>\n\n<\/section>\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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-c2fde16 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"c2fde16\" 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=\"has_eae_slider elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-b9cdda1\" data-id=\"b9cdda1\" 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-73dd69c elementor-widget elementor-widget-text-editor\" data-id=\"73dd69c\" 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 id=\"21b5\" data-selectable-paragraph=\"\">Right-clicking the training node and selecting \u201cRun\u201d will train our model on the training dataset.<\/p>\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-32c54b1 elementor-widget elementor-widget-text-editor\" data-id=\"32c54b1\" 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 id=\"a7b0\" data-selectable-paragraph=\"\">The steps I\u2019ll follow are the following:<\/p>\n\n<ul>\n \t<li id=\"8a56\" data-selectable-paragraph=\"\">Data import<\/li>\n \t<li id=\"41de\" data-selectable-paragraph=\"\">Basic data analysis<\/li>\n \t<li id=\"1346\" data-selectable-paragraph=\"\">Training\/test split<\/li>\n \t<li id=\"d6c9\" data-selectable-paragraph=\"\">Data normalization<\/li>\n \t<li id=\"bbde\" data-selectable-paragraph=\"\">Model selection with cross-validation<\/li>\n \t<li id=\"d7da\" data-selectable-paragraph=\"\">Training of the best model<\/li>\n \t<li id=\"f3e6\" data-selectable-paragraph=\"\">Scoring on the holdout dataset<\/li>\n \t<li id=\"fafe\" data-selectable-paragraph=\"\">Evaluation of the performance<\/li>\n<\/ul>\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-6b05ebb elementor-widget elementor-widget-heading\" data-id=\"6b05ebb\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"87c6\" data-selectable-paragraph=\"\">Free subscription<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-8e84f9f elementor-widget elementor-widget-text-editor\" data-id=\"8e84f9f\" 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 id=\"a810\" data-selectable-paragraph=\"\">At the day I\u2019m writing this article, Azure ML Studio comes with a\u00a0<strong>free subscription\u00a0<\/strong>and many paid subscriptions based on API usage or disk storage. The free subscription comes with 10 GB of disk space and is enough for educational purposes or small-sized experiments. By the way, \u201cExperiment\u201d is the name that ML studio uses to identify a\u00a0<strong>visual workflow<\/strong>. It\u2019s not the only thing we can do with this software since it comes with the well known\u00a0<strong>Jupyter notebooks\u00a0<\/strong>as well. However, in this article, I\u2019ll cover only the visual part of ML studio, i.e. the experiments.<\/p>\n<p id=\"0e39\" data-selectable-paragraph=\"\">In order to create a free subscription, you can visit the URL\u00a0<a href=\"https:\/\/azure.microsoft.com\/en-us\/services\/machine-learning-studio\/\" target=\"_blank\" rel=\"noopener noreferrer\">https:\/\/azure.microsoft.com\/en-us\/services\/machine-learning-studio\/<\/a>, click the \u201cGet started now\u201d button and then select the \u201cFree worskspace\u201d plan. If you already have a Microsoft account (for example, if you have ever used Skype), you can attach an ML Studio subscription to this account; otherwise, you\u2019ll need to create one.<\/p>\n\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-6de867f elementor-widget elementor-widget-heading\" data-id=\"6de867f\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"4eaf\" data-selectable-paragraph=\"\">First approach<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-3e0811f elementor-widget elementor-widget-text-editor\" data-id=\"3e0811f\" 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 id=\"d4f3\" data-selectable-paragraph=\"\">After you\u2019ve finished the subscription procedure, the first window you\u2019ll see opening ML Studio is the following:<\/p>\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-19a086e elementor-widget elementor-widget-image\" data-id=\"19a086e\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1366\/1*It2XGzIGmuF7iORsszYrtw.png\" alt=\"\" \/>\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-d6068e3 elementor-widget elementor-widget-text-editor\" data-id=\"d6068e3\" 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 id=\"be5b\" data-selectable-paragraph=\"\">On the left sidebar, we can find many useful features of ML studio, but in this article, I\u2019ll cover only the\u00a0<strong>experiments\u00a0<\/strong>part.<\/p>\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-c18f53f elementor-widget elementor-widget-text-editor\" data-id=\"c18f53f\" 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 id=\"34c2\" data-selectable-paragraph=\"\">Clicking on the \u201cExperiments\u201d button and then on the \u201cNew\u201d button, we can then create a \u201cBlank experiment\u201d or load a pre-defined one from the\u00a0<strong>gallery<\/strong>.<\/p>\n<p id=\"69c4\" data-selectable-paragraph=\"\">The main screen we are going to work with most of the time is the following:<\/p>\n\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-08044be elementor-widget elementor-widget-image\" data-id=\"08044be\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/798\/1*qM9iRdDyH1T9xBeceljcWA.png\" alt=\"\" \/>\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-1c6640c elementor-widget elementor-widget-text-editor\" data-id=\"1c6640c\" 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 id=\"d910\" data-selectable-paragraph=\"\">The left column contains all the controls that can be dragged and dropped in the central part. The right sidebar is related to the parameters and options of the nodes.<\/p>\n<p id=\"33fe\" data-selectable-paragraph=\"\">Now, we can start with the interesting stuff.<\/p>\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-3a85bc3 elementor-widget elementor-widget-heading\" data-id=\"3a85bc3\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"5ddf\" data-selectable-paragraph=\"\">Data import<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-560278e elementor-widget elementor-widget-text-editor\" data-id=\"560278e\" 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 id=\"5ed4\" data-selectable-paragraph=\"\">ML studio can handle data coming from\u00a0<strong>different sources<\/strong>. It\u2019s possible to upload a dataset from a flat file or reading it from an Azure SQL Database, an URL or even a blob in an Azure Storage Account. The only thing you have to keep in mind is that ML studio supports only a limited number of formats, including CSV and TSV (tab separated values). Concerning the separator, you have only a few fixed options among which you can choose, so be careful when you create a dataset; first, make sure you use a format that ML Studio recognizes.<\/p>\n<p id=\"ca66\" data-selectable-paragraph=\"\">For this simple example, I\u2019ll use the famous Iris dataset. ML Studio contains many\u00a0<strong>example datasets<\/strong>, including a modified version of the Iris dataset, suitable for a binary classification problem.<\/p>\n<p id=\"ab2f\" data-selectable-paragraph=\"\">If you go to \u201cSaved Datasets\u201d and then on \u201cSamples\u201d, you\u2019ll find all the available example sets.<\/p>\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-b2f007e elementor-widget elementor-widget-image\" data-id=\"b2f007e\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/336\/1*BkgN1XYsETZmiz2pT9vBZQ.png\" alt=\"\" \/>\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-87af0b9 elementor-widget elementor-widget-text-editor\" data-id=\"87af0b9\" 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 id=\"6612\" data-selectable-paragraph=\"\">Search for \u201cIris two class data\u201d and drag it on the central part of the screen.<\/p>\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-af39fba elementor-widget elementor-widget-image\" data-id=\"af39fba\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/384\/1*M0yari5UcQ9h1jemknofMQ.png\" alt=\"\" \/>\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-7169e4f elementor-widget elementor-widget-text-editor\" data-id=\"7169e4f\" 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 id=\"c0c4\" data-selectable-paragraph=\"\">The circle with the number 1 in the bottom side of the node is an\u00a0<strong>output port<\/strong>. In ML Studio, each node can have several input and output ports, identified by circles and number. The input ports are located on the upper side of the node and specify the input data the node has to\u00a0<strong>manipulate<\/strong>. The output ports are located at the bottom and are used to\u00a0<strong>distribute<\/strong>\u00a0the output of the node as an input to other nodes. Connecting the nodes through their ports makes us design a\u00a0<strong>complete workflow<\/strong>.<\/p>\n<p id=\"f49e\" data-selectable-paragraph=\"\">Datasets don\u2019t have input port because they don\u2019t crunch data of any kind (they only deliver it).<\/p>\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-6b72f40 elementor-widget elementor-widget-heading\" data-id=\"6b72f40\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"5cdf\" data-selectable-paragraph=\"\">Basic data analysis<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d1715f9 elementor-widget elementor-widget-text-editor\" data-id=\"d1715f9\" 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 id=\"e3d3\" data-selectable-paragraph=\"\">Before manipulating this dataset in any way, we can\u00a0<strong>take a look<\/strong>\u00a0at it. Let\u2019s right-click the node and select \u201cDataset\u201d, then \u201cVisualize\u201d.<\/p>\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-6836f71 elementor-widget elementor-widget-image\" data-id=\"6836f71\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/754\/1*58Ri7VhN5TFjhSHzbQMmCg.png\" alt=\"\" \/>\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-b67e0ea elementor-widget elementor-widget-text-editor\" data-id=\"b67e0ea\" 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 id=\"fce4\" data-selectable-paragraph=\"\">This is the window that appears:<\/p>\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-1a380f3 elementor-widget elementor-widget-image\" data-id=\"1a380f3\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1171\/1*wZ3Ozphebk3gfssnPQOLTQ.png\" alt=\"\" \/>\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-59a2e62 elementor-widget elementor-widget-text-editor\" data-id=\"59a2e62\" 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 id=\"a392\" data-selectable-paragraph=\"\">The central part contains a sample of the dataset and some thumbnails of the\u00a0<strong>histograms<\/strong>\u00a0of each column, that can be selected individually.<\/p>\n<p id=\"5bbe\" data-selectable-paragraph=\"\">The right part contains some\u00a0<strong>basic statistics<\/strong>\u00a0about the column you select.<\/p>\n<p id=\"5f33\" data-selectable-paragraph=\"\">If you scroll down, you\u2019ll see a histogram of the selected variable.<\/p>\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-44faa3d elementor-widget elementor-widget-image\" data-id=\"44faa3d\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1161\/1*vmsmI3y-5eD9GVTlKeyvgg.png\" alt=\"\" \/>\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-9916429 elementor-widget elementor-widget-text-editor\" data-id=\"9916429\" 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 id=\"e460\" data-selectable-paragraph=\"\">A useful feature is a possibility to plot one variable\u00a0<strong>against another<\/strong>\u00a0one via the \u201cCompare to\u201d dropdown menu.<\/p>\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-c4ccc64 elementor-widget elementor-widget-image\" data-id=\"c4ccc64\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1163\/1*YFOE5GgbPM5cY0TKa8BQHw.png\" alt=\"\" \/>\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-4d13dd1 elementor-widget elementor-widget-text-editor\" data-id=\"4d13dd1\" 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 id=\"6a85\" data-selectable-paragraph=\"\">As you can see, this plot highlights a strong, visual correlation between the sepal length and the class variables. It\u2019s a very useful piece of information about the importance of this feature.<\/p>\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-2a7cf47 elementor-widget elementor-widget-heading\" data-id=\"2a7cf47\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"c741\" data-selectable-paragraph=\"\">Training\/test split<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7935ee3 elementor-widget elementor-widget-text-editor\" data-id=\"7935ee3\" 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 id=\"2e60\" data-selectable-paragraph=\"\">The next thing we have to do with our dataset is deciding which part of it we want to use as the\u00a0<strong>training dataset<\/strong>. The remaining part will be used as a test dataset (sometimes called\u00a0<strong>holdout<\/strong>), which will be used only for the final model evaluation.<\/p>\n<p id=\"6b7e\" data-selectable-paragraph=\"\">We can use the \u201cSplit Data\u201d node and connect its input to the output of the dataset.<\/p>\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-f7a2572 elementor-widget elementor-widget-image\" data-id=\"f7a2572\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1286\/1*CVvaN8RssVI316BBeQAKww.png\" alt=\"\" \/>\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-3975f6b elementor-widget elementor-widget-text-editor\" data-id=\"3975f6b\" 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 id=\"41b7\" data-selectable-paragraph=\"\">This way, we are telling ML studio \u201cUse the output of iris dataset node as an input to the split node\u201d.<\/p>\n<p id=\"955e\" data-selectable-paragraph=\"\">On the right part, you can see the options of the node. We can use a 0.7 ratio for the training dataset, while the remaining 0.3 is used for the test set. The split is performed\u00a0<strong>randomly<\/strong>.<\/p>\n<p id=\"f921\" data-selectable-paragraph=\"\">The two output ports of the split node are, respectively, the training dataset (port number 1) and the test dataset (port number 2).<\/p>\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-7339f19 elementor-widget elementor-widget-heading\" data-id=\"7339f19\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"a752\" data-selectable-paragraph=\"\">Data normalization<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e1175ac elementor-widget elementor-widget-text-editor\" data-id=\"e1175ac\" 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 id=\"6409\" data-selectable-paragraph=\"\">In machine learning, it\u2019s frequent that data must be\u00a0<strong>prepared<\/strong>\u00a0for our model in a proper way. The reasons are many and depend on model nature. Logistic regression and neural networks work quite well when the input variables are scaled between 0 and 1. That\u2019s due to the fact that logistic function saturates easily for input values that are greater than 2 on absolute value, so their importance could be misunderstood by the model. With a\u00a0<strong>0-1 scaling<\/strong>, the minimum value of each variable becomes 0 and the maximum value becomes 1. The other values are scaled proportionally. Don\u2019t forget that scaling of the features is an important part of the\u00a0<strong>pre-processing<\/strong>\u00a0part of a machine learning pipeline.<\/p>\n<p id=\"8b9d\" data-selectable-paragraph=\"\">So, we need to use the \u201cNormalize Data\u201d node. From the options panel, we can choose MinMax (that is, 0\u20131 range).<\/p>\n<p id=\"c912\" data-selectable-paragraph=\"\">The input of the Normalize Data node is the training dataset, so we connect it to the first output port of the split data node.<\/p>\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-64c26f0 elementor-widget elementor-widget-image\" data-id=\"64c26f0\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1285\/1*lDtn8afzOx0Kr3zcIokI7g.png\" alt=\"\" \/>\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-77ec5a1 elementor-widget elementor-widget-text-editor\" data-id=\"77ec5a1\" 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 id=\"9417\" data-selectable-paragraph=\"\">The Normalize Data node has two output ports. The first one is the scaled input dataset, the second one make possible to use the scaling transformation in other datasets. It will soon be useful.<\/p>\n\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-ad91469 elementor-widget elementor-widget-heading\" data-id=\"ad91469\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"dd6c\" data-selectable-paragraph=\"\">Model selection with cross-validation<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e2b5827 elementor-widget elementor-widget-text-editor\" data-id=\"e2b5827\" 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 id=\"d6f2\" data-selectable-paragraph=\"\">Now we have prepared our data for the training phase. We are working with a binary classification problem and, in this example, we\u2019ll work with\u00a0<strong>logistic regression\u00a0<\/strong>and a\u00a0<strong>neural network<\/strong>, choosing the best one among the two models.<\/p>\n\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-602ede6 elementor-widget elementor-widget-text-editor\" data-id=\"602ede6\" 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 id=\"5e42\" data-selectable-paragraph=\"\">For each one of the two models, we\u2019ll perform\u00a0<strong>k-fold cross-validation<\/strong>\u00a0in order to check their average performance on\u00a0<strong>unseen data<\/strong>\u00a0and then select the model that has the highest performances.<\/p>\n<p id=\"f4ac\" data-selectable-paragraph=\"\">Let\u2019s start adding the Logistic Regression node by searching for the word \u201clogistic\u201d.<\/p>\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-5b3c9f5 elementor-widget elementor-widget-image\" data-id=\"5b3c9f5\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/357\/1*24Wp1paBy93RSitEqyczKg.png\" alt=\"\" \/>\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-ab86820 elementor-widget elementor-widget-text-editor\" data-id=\"ab86820\" 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 id=\"1dc9\" data-selectable-paragraph=\"\">We can select the node \u201cTwo-Class Logistic Regression\u201d and drag it into the workspace.<\/p>\n<p id=\"1d24\" data-selectable-paragraph=\"\">Then we can search \u201ccross\u201d and will find the \u201cCross Validate Model\u201d node.<\/p>\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-7fbaa4e elementor-widget elementor-widget-image\" data-id=\"7fbaa4e\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/359\/1*nzkWEW3UqvvwmaVvOJJG5g.png\" alt=\"\" \/>\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-40f9580 elementor-widget elementor-widget-text-editor\" data-id=\"40f9580\" 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 id=\"5955\" data-selectable-paragraph=\"\">We can connect the nodes as shown in the next figure:<\/p>\n\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-4131668 elementor-widget elementor-widget-image\" data-id=\"4131668\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/689\/1*hB7l2fjxPqgXm_1KavjWcQ.png\" alt=\"\" \/>\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-3a40e7e elementor-widget elementor-widget-text-editor\" data-id=\"3a40e7e\" 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 id=\"70fb\" data-selectable-paragraph=\"\">On the right part, we must select the target variable:<\/p>\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-41ab30d elementor-widget elementor-widget-image\" data-id=\"41ab30d\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/458\/1*oS65K-V5Tl0CY1e7Thdgkg.png\" alt=\"\" \/>\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-71f43a6 elementor-widget elementor-widget-text-editor\" data-id=\"71f43a6\" 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 id=\"035d\" data-selectable-paragraph=\"\">Click on \u201cLaunch column selector\u201d and select the \u201cClass\u201d variable, as shown in the next picture.<\/p>\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-950210a elementor-widget elementor-widget-image\" data-id=\"950210a\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/899\/1*NpQuOOYVkSAncHgpY7Ztyg.png\" alt=\"\" \/>\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-6748b9c elementor-widget elementor-widget-text-editor\" data-id=\"6748b9c\" 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 id=\"2036\" data-selectable-paragraph=\"\">Now we can run the cross-validation process by right-clicking the \u201cCross Validate Model\u201d node and selecting \u201cRun selected\u201d.<\/p>\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-fbfb82f elementor-widget elementor-widget-image\" data-id=\"fbfb82f\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/695\/1*CZ_-NywFOm-Wu-zVeASGaw.png\" alt=\"\" \/>\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-cd63cfb elementor-widget elementor-widget-text-editor\" data-id=\"cd63cfb\" 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 id=\"45aa\" data-selectable-paragraph=\"\">After the process has ended, we can right click again and select \u201cEvaluation results by fold\u201d, then \u201cVisualize\u201d.<\/p>\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-b9c3fd3 elementor-widget elementor-widget-image\" data-id=\"b9c3fd3\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/750\/1*9zJt8VyJsvJ6dAT2hRZV0Q.png\" alt=\"\" \/>\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-8a85cc0 elementor-widget elementor-widget-text-editor\" data-id=\"8a85cc0\" 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 id=\"00d1\" data-selectable-paragraph=\"\">The following picture shows the evaluation performance metric for each one of the 10 default folds for cross-validation. We\u2019ll check the\u00a0<strong>Area under the ROC curve<\/strong>\u00a0(often called AUC)\u00a0as\u00a0a\u00a0metric\u00a0to\u00a0compare\u00a0different\u00a0models. The higher this value, the better the model.<\/p>\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-d0bccdf elementor-widget elementor-widget-image\" data-id=\"d0bccdf\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1039\/1*QlGC1iGBtMq-b-hDYaqr4w.png\" alt=\"\" \/>\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-5f71b9d elementor-widget elementor-widget-text-editor\" data-id=\"5f71b9d\" 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 id=\"e609\" data-selectable-paragraph=\"\">Scrolling down we\u2019ll reach the \u201cMean\u201d row, which contains the\u00a0<strong>mean<\/strong>\u00a0values of the performance metrics calculated among the folds.<\/p>\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-d7af27c elementor-widget elementor-widget-image\" data-id=\"d7af27c\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/769\/1*B1qR1Fc1oC-AxTefhN0HEw.png\" alt=\"\" \/>\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-461d47f elementor-widget elementor-widget-text-editor\" data-id=\"461d47f\" 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 id=\"e6e0\" data-selectable-paragraph=\"\">The mean value of Logistic Regression\u2019s AUC is 0.9. Let\u2019s keep it in mind.<\/p>\n<p id=\"9e77\" data-selectable-paragraph=\"\">Now it\u2019s time for the neural network, so we\u2019ll repeat the process and search for the word \u201cneural\u201d in the search box.<\/p>\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-791d9d3 elementor-widget elementor-widget-image\" data-id=\"791d9d3\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/352\/1*nedQJRQQRGqxfb67WL2EjA.png\" alt=\"\" \/>\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-0227802 elementor-widget elementor-widget-text-editor\" data-id=\"0227802\" 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 id=\"0688\" data-selectable-paragraph=\"\">We need the \u201cTwo-Class neural network\u201d node, so let\u2019s drag it together with the cross-validation node as the following picture.<\/p>\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-05e7cd2 elementor-widget elementor-widget-image\" data-id=\"05e7cd2\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/621\/1*36tEPzdrkp2huhB6Ebm2Kg.png\" alt=\"\" \/>\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-39e5f9d elementor-widget elementor-widget-text-editor\" data-id=\"39e5f9d\" 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 id=\"2679\" data-selectable-paragraph=\"\">Neural networks have\u00a0<strong>many hyperparameters<\/strong>, so we must choose at least how many neurons we want to use in the\u00a0<strong>hidden layer<\/strong>. For this example, we\u2019ll select 5 hidden nodes.<\/p>\n<p id=\"2e75\" data-selectable-paragraph=\"\">Click on the \u201cTwo-Class Neural Network\u201d node and change the \u201cNumber of hidden nodes\u201d to 5.<\/p>\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-b2a4773 elementor-widget elementor-widget-image\" data-id=\"b2a4773\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/677\/1*RqA0S86VHSMlv8L4p-ohkA.png\" alt=\"\" \/>\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-3cec020 elementor-widget elementor-widget-text-editor\" data-id=\"3cec020\" 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 id=\"345e\" data-selectable-paragraph=\"\">As you can see, it\u2019s equal to the logistic regression performance. It\u2019s due to the nature of iris dataset, which is chosen to make every model work properly.<\/p>\n<p id=\"3251\" data-selectable-paragraph=\"\">If one of the two models would have reached higher performances than the other one, we would have selected it. Since the performances are the same and we want the\u00a0<strong>simplest model possible<\/strong>, we\u2019ll choose the logistic regression.<\/p>\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-44a9e30 elementor-widget elementor-widget-heading\" data-id=\"44a9e30\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"54ab\" data-selectable-paragraph=\"\">Training the best model<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e962192 elementor-widget elementor-widget-text-editor\" data-id=\"e962192\" 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 id=\"3c24\" data-selectable-paragraph=\"\">Now we can safely train the logistic regression over the\u00a0<strong>entire training dataset<\/strong>\u00a0since cross-validation has shown that training this model doesn\u2019t introduce biases or overfitting.<\/p>\n<p id=\"9974\" data-selectable-paragraph=\"\">Training a model on a dataset can be done using the \u201cTrain model\u201d node. The first input port is the model itself, while the second input port is the training dataset.<\/p>\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-3b2adb2 elementor-widget elementor-widget-image\" data-id=\"3b2adb2\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/339\/1*VJrUMl8LMoFNtah706xQDg.png\" alt=\"\" \/>\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<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-4521e6d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"4521e6d\" 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=\"has_eae_slider elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-891bcd8\" data-id=\"891bcd8\" 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-f31e5ac elementor-widget elementor-widget-text-editor\" data-id=\"f31e5ac\" 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 id=\"65cc\" data-selectable-paragraph=\"\">Clicking on the \u201cTrain model\u201d node, we are allowed to select the target column, which is still the \u201cClass\u201d variable.<\/p>\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-36be27f elementor-widget elementor-widget-image\" data-id=\"36be27f\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/901\/1*tHZrMxQHYpTAd-MtKmFMCg.png\" alt=\"\" \/>\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-adba148 elementor-widget elementor-widget-text-editor\" data-id=\"adba148\" 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 id=\"21b5\" data-selectable-paragraph=\"\">Right-clicking the training node and selecting \u201cRun\u201d will train our model on the training dataset.<\/p>\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-7874249 elementor-widget elementor-widget-heading\" data-id=\"7874249\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"9d8e\" data-selectable-paragraph=\"\">Scoring the holdout dataset<\/h1><\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1492c78 elementor-widget elementor-widget-text-editor\" data-id=\"1492c78\" 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 id=\"d773\" data-selectable-paragraph=\"\">The next thing we have to do is apply our model on the\u00a0<strong>holdout\u00a0<\/strong>dataset in order to quantify how the model performs on data it has never seen during training.<\/p>\n<p id=\"9ec9\" data-selectable-paragraph=\"\">Remember, we have previously scaled the training dataset, so we have to perform the\u00a0<strong>same transformation<\/strong>\u00a0on the holdout in order to make the model work properly.<\/p>\n<p id=\"e0e9\" data-selectable-paragraph=\"\">Applying a previous transformation to a dataset is possible using the \u201cApply Transformation\u201d node.<\/p>\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-af0ec90 elementor-widget elementor-widget-image\" data-id=\"af0ec90\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/356\/1*9047ePNsar3mkQrMrr9CAg.png\" alt=\"\" \/>\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-4c8ad14 elementor-widget elementor-widget-text-editor\" data-id=\"4c8ad14\" 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 id=\"efbe\" data-selectable-paragraph=\"\">Remember the second output port of the \u201cNormalize data\u201d node? It\u2019s time to connect it to the first input port of the \u201cApply transformation node\u201d. The second input port is the holdout dataset, which is the last output port of the split data node.<\/p>\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-8d16bbe elementor-widget elementor-widget-image\" data-id=\"8d16bbe\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/464\/1*caCKxPK_4YQWuDbrnPbhqQ.png\" alt=\"\" \/>\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-3dd6570 elementor-widget elementor-widget-text-editor\" data-id=\"3dd6570\" 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 id=\"19c1\" data-selectable-paragraph=\"\">This way, we are telling ML Studio to apply, to the holdout dataset, the same normalize transform used for the training dataset. This is very important because our model has been trained on transformed data and the same transformation must be used in\u00a0<strong>every dataset<\/strong>\u00a0we want our model to score.<\/p>\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-bcf5bff elementor-widget elementor-widget-heading\" data-id=\"bcf5bff\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"8162\" data-selectable-paragraph=\"\">Evaluate the performance<\/h1>\n<\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-206c254 elementor-widget elementor-widget-text-editor\" data-id=\"206c254\" 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 id=\"9f20\" data-selectable-paragraph=\"\">In order to calculate the performances of our model in the holdout, we must make the\u00a0<strong>scoring\u00a0<\/strong>of the dataset. The operation of giving a dataset to a model is called \u201cScoring\u201d. The model takes the dataset and returns its prediction, which is a probability that the event labeled with 1 occurs. This probability (called \u201cscore\u201d), compared with the real occurring events in the holdout dataset (which the model doesn\u2019t know) will make us evaluate model performance.<\/p>\n<p id=\"7b93\" data-selectable-paragraph=\"\">In order to score the dataset, we can search for the \u201cScore model\u201d node.<\/p>\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-b6edee5 elementor-widget elementor-widget-image\" data-id=\"b6edee5\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/351\/1*zEHG-WelTAxHFlRBFHRTlA.png\" alt=\"\" \/>\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-7558f71 elementor-widget elementor-widget-text-editor\" data-id=\"7558f71\" 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 id=\"d7c6\" data-selectable-paragraph=\"\">Then, we can add it to the workflow in this way.<\/p>\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-13e6b30 elementor-widget elementor-widget-image\" data-id=\"13e6b30\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/496\/1*iNXCAbP6Y2GDZJSJnnEKXg.png\" alt=\"\" \/>\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-0a61fb8 elementor-widget elementor-widget-text-editor\" data-id=\"0a61fb8\" 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 id=\"e31a\" data-selectable-paragraph=\"\">The first input port is the trained model (i.e. the output of the Train model node), while the second input port is the dataset to score (in our case, the transformed holdout dataset).<\/p>\n<p id=\"2fdb\" data-selectable-paragraph=\"\">After executing the score model node, we can take a look at what it does.<\/p>\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-75b5f20 elementor-widget elementor-widget-image\" data-id=\"75b5f20\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/643\/1*b4CYBiPmfp6zcjnAgq0iwA.png\" alt=\"\" \/>\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-2d56204 elementor-widget elementor-widget-text-editor\" data-id=\"2d56204\" 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 id=\"4386\" data-selectable-paragraph=\"\">As you can see, there is a new pair of columns called \u201cScored Labels\u201d and \u201cScored Probabilities\u201d. The second one is the probability that the target label is 1, while the first one is the predicted target itself, calculates as 1 if the probability is greater than 50% and 0 otherwise.<\/p>\n\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-a769517 elementor-widget elementor-widget-image\" data-id=\"a769517\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/720\/1*ijFXMXxIq_iyrxrYOaAmDw.png\" alt=\"\" \/>\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-64ea92d elementor-widget elementor-widget-text-editor\" data-id=\"64ea92d\" 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 id=\"e86b\" data-selectable-paragraph=\"\">Finally, we can use the \u201cEvaluate Model\u201d node to extract the performance metrics we need.<\/p>\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-cabd729 elementor-widget elementor-widget-image\" data-id=\"cabd729\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/356\/1*s--qi_AmhM254NSLcjgP8g.png\" alt=\"\" \/>\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-9e3e9ac elementor-widget elementor-widget-text-editor\" data-id=\"9e3e9ac\" 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 id=\"8fbb\" data-selectable-paragraph=\"\">We can connect the score model node to the evaluation node and run it.<\/p>\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-b4863c3 elementor-widget elementor-widget-image\" data-id=\"b4863c3\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/264\/1*S9LaoZPrRRQh-d48aUIYJQ.png\" alt=\"\" \/>\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-ff74823 elementor-widget elementor-widget-text-editor\" data-id=\"ff74823\" 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 id=\"90eb\" data-selectable-paragraph=\"\">Finally, these are the results.<\/p>\n\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-9b9399e elementor-widget elementor-widget-image\" data-id=\"9b9399e\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1124\/1*wos6o1pBo82iMv-g7ZQr5A.png\" alt=\"\" \/>\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-51bedd8 elementor-widget elementor-widget-text-editor\" data-id=\"51bedd8\" 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 id=\"dd1d\" data-selectable-paragraph=\"\">The panel on the left is the ROC curve, which is pretty amazing.<\/p>\n<p id=\"a331\" data-selectable-paragraph=\"\">Scrolling down, we can find all the numbers we are looking for.<\/p>\n\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-180b9fb elementor-widget elementor-widget-image\" data-id=\"180b9fb\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/1197\/1*grTVEGpB3B9EgQ1ZyBdK1g.png\" alt=\"\" \/>\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-c70ef86 elementor-widget elementor-widget-text-editor\" data-id=\"c70ef86\" 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 id=\"b6de\" data-selectable-paragraph=\"\">On the upper left part, we have the\u00a0<strong>confusion matrix<\/strong>. Next, we have the\u00a0<strong>standard metrics\u00a0<\/strong>for a binary classification model (Accuracy, Precision and so on). The right slider changes the\u00a0<strong>threshold\u00a0<\/strong>that transforms the probability in the 0\u20131 label. When you change the threshold, all the metrics change automatically, except for the AUC, which is threshold-independent.<\/p>\n<p id=\"ef7a\" data-selectable-paragraph=\"\">We can say that our model is wonderful (high AUC, high precision, high accuracy), so we can save it inside ML Studio to use it in other experiments.<\/p>\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-df1507f elementor-widget elementor-widget-image\" data-id=\"df1507f\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/659\/1*OSDOn0oLuxuwcj9JpPw2ZA.png\" alt=\"\" \/>\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-6d79a2c elementor-widget elementor-widget-image\" data-id=\"6d79a2c\" data-element_type=\"widget\" data-e-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\" src=\"https:\/\/miro.medium.com\/max\/505\/1*XmGryJvq_Gr9oYA4bpmo3Q.png\" alt=\"\" \/>\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-97526b3 elementor-widget elementor-widget-heading\" data-id=\"97526b3\" 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<h1 class=\"elementor-heading-title elementor-size-default\"><h1 id=\"ae1d\" data-selectable-paragraph=\"\">Conclusion<\/h1>\n<\/h1>\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-821526b elementor-widget elementor-widget-text-editor\" data-id=\"821526b\" 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 id=\"9877\" data-selectable-paragraph=\"\">In this short article, I\u2019ve shown a simple example of the use of Azure ML studio. It\u2019s a very useful tool in the machine learning industry and, although it has some limits (limited number of records, limited choice of models), I think that even the most code-oriented data scientist will love this simple tool. It\u2019s pretty worth mentioning that, paying the appropriate fee, ML studio can be used for real-time training and prediction thanks to its strong REST API interface. This enables many possible machine learning scenarios.<\/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>In this short article, a simple example of the use of Azure ML studio is shown. It&rsquo;s a very useful tool in the machine learning industry and, although it has some limits with limited number of records, limited choice of models. Even the most code-oriented data scientist will love this simple tool. It&rsquo;s pretty worth mentioning that, paying the appropriate fee, ML studio can be used for real-time training and prediction thanks to its strong REST API interface. This enables many possible machine learning scenarios.<\/p>\n","protected":false},"author":618,"featured_media":3804,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[183],"tags":[92],"ppma_author":[3328],"class_list":["post-1923","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-ml","tag-machine-learning"],"authors":[{"term_id":3328,"user_id":618,"is_guest":0,"slug":"gianluca-malato","display_name":"Gianluca Malato","avatar_url":"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/04\/medium_918623b2-8f36-4110-8343-6fc9228595dd-150x150.jpg","user_url":"http:\/\/www.gianlucamalato.it\/","last_name":"Malato","first_name":"Gianluca","job_title":"","description":"Gianluca Malato is Data Scientist at Poste Italiane SPA.\u00a0 He is also a fiction author and software developer, Editor of\u00a0<a href=\"https:\/\/medium.com\/data-science-journal?source=follow_footer--------------------------follow_footer-\">Data Science Journal<\/a>,\u00a0<a href=\"https:\/\/medium.com\/the-trading-scientist?source=follow_footer--------------------------follow_footer-\">The Trading Scientist<\/a>, and\u00a0<a href=\"https:\/\/medium.com\/the-writers-notebook?source=follow_footer--------------------------follow_footer-\">The Writer\u2019s Notebook<\/a>. His books are available on <a href=\"https:\/\/www.amazon.com\/Gianluca-Malato\/e\/B076CHTG3W?ref=dbs_a_mng_rwt_scns_share\">Amazon<\/a>."}],"_links":{"self":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/1923","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/users\/618"}],"replies":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/comments?post=1923"}],"version-history":[{"count":4,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/1923\/revisions"}],"predecessor-version":[{"id":36696,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/1923\/revisions\/36696"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/media\/3804"}],"wp:attachment":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/media?parent=1923"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/categories?post=1923"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/tags?post=1923"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/ppma_author?post=1923"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}