{"id":14130,"date":"2020-11-11T10:41:51","date_gmt":"2020-11-11T10:41:51","guid":{"rendered":"https:\/\/www.experfy.com\/blog\/?p=11209"},"modified":"2023-10-05T13:29:45","modified_gmt":"2023-10-05T13:29:45","slug":"stakeholders-interpretable-machine-learning-model","status":"publish","type":"post","link":"https:\/\/www.experfy.com\/blog\/ai-ml\/stakeholders-interpretable-machine-learning-model\/","title":{"rendered":"So, Your Stakeholders Want An Interpretable Machine Learning Model?"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"14130\" class=\"elementor elementor-14130\" 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-465fa38b elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"465fa38b\" 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-11c1b45c\" data-id=\"11c1b45c\" 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\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-b0ffe65 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"b0ffe65\" 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-54dffc4\" data-id=\"54dffc4\" 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-8a3b21c elementor-widget elementor-widget-heading\" data-id=\"8a3b21c\" 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\">Setting the scene<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-8c3f54a elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"8c3f54a\" 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-d13ecfa\" data-id=\"d13ecfa\" 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-37cccb4 elementor-widget elementor-widget-text-editor\" data-id=\"37cccb4\" 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>You are a Data Scientist working for a commercial company. You spent weeks, or maybe even months, developing this deep learning-based model that accurately predicts an outcome of great interest to your business. You proudly presented the results to your stakeholders. Quite annoyingly, though, they did not pay much attention to that cutting-edge approach you used to <a href=\"https:\/\/www.experfy.com\/blog\/ai-ml\/how-to-build-a-machine-learning-model\/\" target=\"_blank\" rel=\"noreferrer noopener\">build the model<\/a>. Instead of focusing on how powerful the model was, they started asking lots of questions on\u00a0<em>why<\/em>\u00a0some of its predictions looked the way they did. Your colleagues also felt that some of the critical predictors were missing. They could not fully understand how the predictions were so accurate with those features missing. As the model you built was of a black box type, it was challenging for you to give satisfactory answers to all the questions straightaway. So you had to ask for a follow-up meeting and for some time to get prepared.<\/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-49c7529 elementor-widget elementor-widget-text-editor\" data-id=\"49c7529\" 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>Sounds familiar? I have certainly been there a few times before. It is natural for us as humans to be uncomfortable with and do not trust things we do not understand. This also applies to the Machine Learning model and how people who are not Data Science experts perceive them. However, having an interpretable Machine Learning model is neither always possible nor necessary. To help me explain that to my stakeholders and clients, I have collected a few key ideas on the topic of model interpretability from various sources, including my own experience. I am sharing this collection here in this article. Hopefully, some of my fellow Data Scientists will also find it useful when preparing for similar conversations with their colleagues.<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-023f9f8 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"023f9f8\" 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-060f9c7\" data-id=\"060f9c7\" 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-e118c75 elementor-widget elementor-widget-heading\" data-id=\"e118c75\" 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\">Be crystal clear about the model\u2019s purpose<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-6d99a04 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6d99a04\" 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-cf21926\" data-id=\"cf21926\" 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-823ba60 elementor-widget elementor-widget-text-editor\" data-id=\"823ba60\" 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>Clearly communicating the purpose of a model is one of the crucial factors that drive its adoption by stakeholders. There are several classifications of model purposes (e.g.,\u00a0<a href=\"https:\/\/royalsocietypublishing.org\/doi\/10.1098\/rsos.172096\" target=\"_blank\" rel=\"noreferrer noopener\" class=\"broken_link\">Calder et al. 2018<\/a>;\u00a0<a href=\"http:\/\/jasss.soc.surrey.ac.uk\/22\/3\/6.html\" target=\"_blank\" rel=\"noreferrer noopener\"> Edmonds et al. 2018<\/a>;\u00a0<a href=\"https:\/\/www.nature.com\/articles\/s41467-020-17785-2\" target=\"_blank\" rel=\"noreferrer noopener\">Grimm et al. 2020<\/a>). I personally prefer the one proposed by\u00a0<a href=\"https:\/\/en.wikipedia.org\/wiki\/Leo_Breiman\" target=\"_blank\" rel=\"noreferrer noopener\">Leo Breiman<\/a>, author of the famous Random Forest algorithm. In 2001, Breiman published a paper entitled \u201c<a href=\"https:\/\/projecteuclid.org\/download\/pdf_1\/euclid.ss\/1009213726\" target=\"_blank\" rel=\"noreferrer noopener\">Statistical Modeling: The Two Cultures<\/a>\u201d. This paper has received lots of interest and citations as it for the first time initiated a widespread discussion on model interpretability vs predictive performance<\/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-8b5f5f9 elementor-widget elementor-widget-text-editor\" data-id=\"8b5f5f9\" 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>According to Breiman, data can be thought of as being generated by a black box, inside which Nature links a vector of input variables\u00a0<strong><em>x<\/em><\/strong>\u00a0to the outcomes\u00a0<strong><em>y<\/em><\/strong><\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-6413558 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6413558\" 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-c21458e\" data-id=\"c21458e\" 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-df37fb8 elementor-widget elementor-widget-image\" data-id=\"df37fb8\" 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 fetchpriority=\"high\" decoding=\"async\" width=\"769\" height=\"177\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_FN1UCVgTRJBz8SJMqQLGEA.png\" class=\"attachment-large size-large wp-image-33267\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_FN1UCVgTRJBz8SJMqQLGEA.png 769w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_FN1UCVgTRJBz8SJMqQLGEA-300x69.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_FN1UCVgTRJBz8SJMqQLGEA-610x140.png 610w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_FN1UCVgTRJBz8SJMqQLGEA-750x173.png 750w\" sizes=\"(max-width: 769px) 100vw, 769px\" \/>\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-9292ba5 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"9292ba5\" 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-879bc20\" data-id=\"879bc20\" 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-6b572e1 elementor-widget elementor-widget-text-editor\" data-id=\"6b572e1\" 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>The author then formulates two main goals of modelling:<\/p>\n\n\n<ul class=\"wp-block-list\">\n<li><em>information:<\/em>\u00a0extracting insights on how Nature is linking\u00a0<strong><em>x<\/em><\/strong>\u00a0to\u00a0<strong><em>y<\/em><\/strong>;<\/li>\n<li><em>prediction:\u00a0<\/em>providing accurate estimates of\u00a0<strong><em>y<\/em><\/strong>\u00a0based on the future values of\u00a0<strong><em>x<\/em><\/strong><em>.<\/em><\/li>\n<\/ul>\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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-fa43c3d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"fa43c3d\" 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-2e55dca\" data-id=\"2e55dca\" 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-fd22b5d elementor-widget elementor-widget-text-editor\" data-id=\"fd22b5d\" 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>When communicating with the model end-users, it is essential to differentiate between these two goals. The reason is that the models used to achieve these goals typically differ in terms of their complexity, interpretability, and predictive power.<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-80ba169 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"80ba169\" 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-1bf149a\" data-id=\"1bf149a\" 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-94c488c elementor-widget elementor-widget-heading\" data-id=\"94c488c\" 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\">Articulate the trade-off between interpretability and predictive accuracy<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-babb972 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"babb972\" 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-fd643c9\" data-id=\"fd643c9\" 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-696439e elementor-widget elementor-widget-text-editor\" data-id=\"696439e\" 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>Breiman (2001) distinguishes two approaches, or \u201ccultures\u201d, toward the goals of modelling.<\/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-2f7b4e4 elementor-widget elementor-widget-text-editor\" data-id=\"2f7b4e4\" 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>The first approach,\u00a0<em>\u201cdata modelling\u201d<\/em>, assumes that the data-generating process can be described by a stochastic model, e.g.\u00a0<em>response = f(predictors, parameters, random noise).<\/em>Such models tend to have a limited set of parameters, whose values are estimated from the observed data. Examples include linear regression, logistic regression, Cox regression, etc.:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-dda54ef elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"dda54ef\" 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-c45fe74\" data-id=\"c45fe74\" 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-af28941 elementor-widget elementor-widget-image\" data-id=\"af28941\" 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\" width=\"1018\" height=\"262\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_y7AB8Y8wDpSGMtDUVdQjMA.png\" class=\"attachment-large size-large wp-image-33268\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_y7AB8Y8wDpSGMtDUVdQjMA.png 1018w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_y7AB8Y8wDpSGMtDUVdQjMA-300x77.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_y7AB8Y8wDpSGMtDUVdQjMA-768x198.png 768w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_y7AB8Y8wDpSGMtDUVdQjMA-610x157.png 610w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_y7AB8Y8wDpSGMtDUVdQjMA-750x193.png 750w\" sizes=\"(max-width: 1018px) 100vw, 1018px\" \/>\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-e1357bc elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e1357bc\" 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-37e0e7c\" data-id=\"37e0e7c\" 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-ee906ea elementor-widget elementor-widget-text-editor\" data-id=\"ee906ea\" 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>Due to their relatively simple structure, models from this first category are typically used to shed light on how the system of interest operates. For instance, one can directly look at the coefficients of a linear regression model and quickly work out how changing the input values will affect the response variable. This also helps with formulating hypotheses that can subsequently be tested in controlled experiments. Although these models can certainly be used to make predictions, the quality of such predictions is usually\u00a0<em> not<\/em> that high.<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-831a250 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"831a250\" 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-21da031\" data-id=\"21da031\" 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-cdb8710 elementor-widget elementor-widget-text-editor\" data-id=\"cdb8710\" 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>This is in contrast to models produced using the\u00a0<em>\u201calgorithmic modelling\u201d<\/em>\u00a0approach. This approach accepts the fact that the innards of Nature\u2019s black box are complex and unknown. It then tries to find an arbitrarily complex function that provides accurate mapping of the input variables\u00a0<strong><em>x<\/em><\/strong>\u00a0to the response variables\u00a0<strong><em>y<\/em><\/strong>. Models that belong to this category are typically more complex and fitted using such algorithms as Random Forest, XGBoost, neural nets, etc.:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-ec7eb22 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"ec7eb22\" 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-ccb0b3a\" data-id=\"ccb0b3a\" 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-7ca668e elementor-widget elementor-widget-image\" data-id=\"7ca668e\" 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\" width=\"974\" height=\"501\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_39c-vGcIHLenT57DrFYigg.png\" class=\"attachment-large size-large wp-image-33269\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_39c-vGcIHLenT57DrFYigg.png 974w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_39c-vGcIHLenT57DrFYigg-300x154.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_39c-vGcIHLenT57DrFYigg-768x395.png 768w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_39c-vGcIHLenT57DrFYigg-610x314.png 610w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_39c-vGcIHLenT57DrFYigg-750x386.png 750w\" sizes=\"(max-width: 974px) 100vw, 974px\" \/>\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-e1eab64 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"e1eab64\" 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-50cfb69\" data-id=\"50cfb69\" 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-feef289 elementor-widget elementor-widget-text-editor\" data-id=\"feef289\" 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>Quality of the models produced with the data modelling approach is usually evaluated using statistical tests for goodness-of-fit and by examining the residuals. The result of this analysis is often binary: the model is either considered \u201cgood\u201d or discarded as a \u201cbad\u201d one. In contrast, models built using the algorithmic modelling approach are assessed based on the accuracy of their predictions on an independent dataset. This is an important distinction, as it implies that we do\u00a0<em>not<\/em>\u00a0really care how complex an algorithmic model is or whether it passes statistical tests for goodness-of-fit. All that matters is that the model does not overfit and its predictive power is sufficiently high for the problem at hand.<\/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-0cf1cdb elementor-widget elementor-widget-text-editor\" data-id=\"0cf1cdb\" 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>Nowadays, businesses collect large volumes of increasingly more complex data. Solving real-world business problems that require high-quality predictions based on such data requires equally complex modelling. However, complex models are intrinsically more difficult to interpret. Although this trade-off is not always black-and-white, we can conceptually visualise it as follows:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-cbc2e58 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"cbc2e58\" 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-3563b9a\" data-id=\"3563b9a\" 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-349159f elementor-widget elementor-widget-image\" data-id=\"349159f\" 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 loading=\"lazy\" decoding=\"async\" width=\"880\" height=\"264\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_QJpYaFIm6xRvBIIKy8Uv2Q.png\" class=\"attachment-large size-large wp-image-33271\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_QJpYaFIm6xRvBIIKy8Uv2Q.png 880w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_QJpYaFIm6xRvBIIKy8Uv2Q-300x90.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_QJpYaFIm6xRvBIIKy8Uv2Q-768x230.png 768w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_QJpYaFIm6xRvBIIKy8Uv2Q-610x183.png 610w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_QJpYaFIm6xRvBIIKy8Uv2Q-750x225.png 750w\" sizes=\"(max-width: 880px) 100vw, 880px\" \/>\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-61fab5e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"61fab5e\" 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-987a20f\" data-id=\"987a20f\" 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-9517c80 elementor-widget elementor-widget-text-editor\" data-id=\"9517c80\" 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>It is our job as Data Scientists to articulate this trade-off to the end-users of our models. It can be challenging to do. However, as\u00a0<a href=\"https:\/\/medium.com\/u\/2fccb851bb5e?source=post_page-----6b13928892de--------------------------------\" target=\"_blank\" rel=\"noreferrer noopener\" class=\"broken_link\">Cassie Kozyrkov<\/a>says in her brilliant\u00a0<a href=\"https:\/\/hackernoon.com\/explainable-ai-wont-deliver-here-s-why-6738f54216be\" target=\"_blank\" rel=\"noreferrer noopener\">article<\/a>\u00a0on this topic,\u00a0<em>\u201cnot everything in life is simple\u201d<\/em>\u00a0and\u00a0<em>\u201cwishing complicated things were simple does not make them so.\u201d<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-1a4665d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"1a4665d\" 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-d50c190\" data-id=\"d50c190\" 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-216caac elementor-widget elementor-widget-heading\" data-id=\"216caac\" 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\">How about LIME, SHAP, and other methods that \u201cexplain\u201d black box models?<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-fcde8a4 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"fcde8a4\" 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-e19d67a\" data-id=\"e19d67a\" 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-6df2495 elementor-widget elementor-widget-text-editor\" data-id=\"6df2495\" 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>Interpretable Machine Learning (a.k.a. Explainable AI, XAI) is definitely a hot topic these days. Many academic researchers, developers of open-source frameworks, and vendors of commercial platforms are churning out novel methods to interpret the inner workings of complex predictive models. Examples of some of the well-known techniques include (see\u00a0<a href=\"https:\/\/christophm.github.io\/interpretable-ml-book\/index.html\" target=\"_blank\" rel=\"noreferrer noopener\">Molnar 2020<\/a>\u00a0for a comprehensive overview):<\/p>\n<p><!-- \/wp:paragraph --><\/p>\n<p><!-- wp:list --><\/p>\n<p>\u00a0<\/p>\n<ul>\n<li style=\"list-style-type: none;\">\n<ul>\n<li>LIME (Local Interpretable Model-Agnostic Explanations);<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u00a0<\/p>\n<ul>\n<li style=\"list-style-type: none;\">\n<ul>\n<li>Shapely values and the associated SHAP method;<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u00a0<\/p>\n<ul>\n<li style=\"list-style-type: none;\">\n<ul>\n<li>partial dependence plot;<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u00a0<\/p>\n<ul>\n<li style=\"list-style-type: none;\">\n<ul>\n<li>feature importance;<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u00a0<\/p>\n<ul>\n<li style=\"list-style-type: none;\">\n<ul>\n<li>individual conditional expectation;<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u00a0<\/p>\n<ul>\n<li style=\"list-style-type: none;\">\n<ul>\n<li>accumulated local effects plot.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u00a0<\/p>\n<p><!-- \/wp:list --><\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-3e704fd elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"3e704fd\" 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-c88619a\" data-id=\"c88619a\" 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-b3afc4e elementor-widget elementor-widget-text-editor\" data-id=\"b3afc4e\" 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>First of all, no Data Science project should be developed in a vacuum. This means that business stakeholders should be involved from Day 1. Before jumping into what we as Data Scientists love most \u2014 model building and playing with algorithms \u2014 we should strive to collect as much domain knowledge from our business colleagues as possible. On the one hand, embedding this knowledge in the form of input features would increase the chance of developing a highly performant model. On the other hand, this would eventually minimise the need to explain how the model works.<\/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-922bc00 elementor-widget elementor-widget-text-editor\" data-id=\"922bc00\" 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>But sometimes we do develop models without much engagement from our stakeholders (e.g., as part of an R&amp;amp;D project). In such cases, I have found it useful to simply provide a detailed explanation of what input variables go into the model under discussion. Business folks will naturally have an intuition as to what variables are likely to drive the outcome of interest. And if they see that these variables are already part of the model, their trust toward it goes up.<\/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-c5a74f7 elementor-widget elementor-widget-text-editor\" data-id=\"c5a74f7\" 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>Another powerful thing that often works well is exposing the model via a simple interactive web application to illustrate how predictions change depending on the input values. This can be done using any of the popular frameworks, such as\u00a0<a href=\"https:\/\/shiny.rstudio.com\/\" target=\"_blank\" rel=\"noreferrer noopener\"> Shiny<\/a>,\u00a0<a href=\"https:\/\/plotly.com\/dash\/\" target=\"_blank\" rel=\"noreferrer noopener\"> Dash<\/a>,\u00a0<a href=\"https:\/\/www.streamlit.io\/\" target=\"_blank\" rel=\"noreferrer noopener\">Streamlit<\/a>. Let your stakeholders move those sliders and run the wildest what-if scenarios! This can dramatically improve their understanding of the model, better than any feature importance plot could do.<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-41a1e8e elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"41a1e8e\" 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-77ddc09\" data-id=\"77ddc09\" 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-3c1c751 elementor-widget elementor-widget-heading\" data-id=\"3c1c751\" 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\">Remind your colleagues that correlation is not causation<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-8ff6bfa elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"8ff6bfa\" 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-a9830d9\" data-id=\"a9830d9\" 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-1318f9f elementor-widget elementor-widget-text-editor\" data-id=\"1318f9f\" 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>Predictive models that capture actual mechanistic links between the input and the response variables are rather rare in business settings. This is especially true for the complex algorithmic models that include a large number of predictors. Most of these models make their predictions merely due to the\u00a0&lt;em&gt;correlation&lt;\/em&gt;\u00a0between predictors and the response variable. But, as the saying goes, \u201c<a href=\"https:\/\/www.theguardian.com\/science\/blog\/2012\/jan\/06\/correlation-causation\" rel=\"noopener\">correlation is not causation<\/a>\u201d, or at least not always. This has two important implications when it comes to interpreting a predictive 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-c835e65 elementor-widget elementor-widget-text-editor\" data-id=\"c835e65\" 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>Firstly, it\u00a0<em>is<\/em>\u00a0possible to build a useful predictive model using input variables that have\u00a0<em> no<\/em>\u00a0<em> actual<\/em>\u00a0association with the response variable. One can find lots of\u00a0<a href=\"https:\/\/tylervigen.com\/spurious-correlations\" target=\"_blank\" rel=\"noreferrer noopener\">examples<\/a>\u00a0on the Internet of the so-called \u201c<a href=\"https:\/\/en.wikipedia.org\/wiki\/Spurious_relationship\" target=\"_blank\" rel=\"noreferrer noopener\"> spurious correlations<\/a>\u201d. Here is one of them:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-73ddf21 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"73ddf21\" 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-f674ea3\" data-id=\"f674ea3\" 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-6bd44a1 elementor-widget elementor-widget-image\" data-id=\"6bd44a1\" 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 loading=\"lazy\" decoding=\"async\" width=\"666\" height=\"360\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_aD239pBdMGEHj37SVA4AXQ.png\" class=\"attachment-large size-large wp-image-33273\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_aD239pBdMGEHj37SVA4AXQ.png 666w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_aD239pBdMGEHj37SVA4AXQ-300x162.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_aD239pBdMGEHj37SVA4AXQ-610x330.png 610w\" sizes=\"(max-width: 666px) 100vw, 666px\" \/>\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-6de6dc7 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6de6dc7\" 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-c4f013e\" data-id=\"c4f013e\" 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-af3622d elementor-widget elementor-widget-text-editor\" data-id=\"af3622d\" 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>It is easy to build a simple linear regression model that would accurately predict the number of doctorate degrees awarded based on mozzarella cheese consumption. Can this model be successfully used in practice to estimate the number of doctorate degrees awarded in a given year? Thanks to the high correlation between the two variables, definitely yes. But something tells me that any attempt to\u00a0<em>interpret<\/em>\u00a0it would only trigger a good laugh in the room.<\/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-9424f42 elementor-widget elementor-widget-text-editor\" data-id=\"9424f42\" 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>Secondly, it is often tempting to use insights gained from a predictive model to devise actions for\u00a0<em> controlling<\/em>\u00a0the response variable. However, if such a model is mainly based on non-mechanistic associations, doing so is likely to be meaningless, and sometimes can be even dangerous. For example, the relationship depicted above implies that increasing the per capita consumption of mozzarella cheese would result in more doctorate degrees awarded. Give it a moment to sink in\u2026 Would you recommend this to a decision maker whose goal is to strengthen the workforce with more civil engineers educated to a PhD level?<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-102ec02 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"102ec02\" 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-23437ef\" data-id=\"23437ef\" 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-dce1091 elementor-widget elementor-widget-heading\" data-id=\"dce1091\" 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\">Explain the existence of multiple good model<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-6433987 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"6433987\" 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-11f47a2\" data-id=\"11f47a2\" 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-dcea758 elementor-widget elementor-widget-text-editor\" data-id=\"dcea758\" 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>By definition, any model is only an approximation of the process that generated the observed data. As this underlying process is unknown, the same data can often be described similarly well by very different models. In Statistics, this phenomenon is known as the\u00a0<em>\u201cmultiplicity of models\u201d<\/em>\u00a0(Breiman 2001).<\/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-3b9422d elementor-widget elementor-widget-text-editor\" data-id=\"3b9422d\" 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>This phenomenon causes no problems as long as all we need from a model is high predictive accuracy. However, it becomes problematic if the goal is to gain <em> insights<\/em>\u00a0about the data-generating process and then make\u00a0<em> practical decisions<\/em>\u00a0based on this information. The reason is simple: different models that fit the same data well can lead to remarkably different conclusions. And the worst part is that there is\u00a0<em>no way<\/em>\u00a0to tell which of these conclusions are correct (unless they are proved correct in a follow-up controlled experiment).<\/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-0f55e8a elementor-widget elementor-widget-text-editor\" data-id=\"0f55e8a\" 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>Here is a simple example. Suppose, we have a dataset that describes how a response variable\u00a0<em>y<\/em>\u00a0changes over time<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-4431e77 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"4431e77\" 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-cc25685\" data-id=\"cc25685\" 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-e6eb54b elementor-widget elementor-widget-image\" data-id=\"e6eb54b\" 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 loading=\"lazy\" decoding=\"async\" width=\"666\" height=\"360\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_Z4EMQNj7ROm7k0LyHaWE_w.png\" class=\"attachment-large size-large wp-image-33274\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_Z4EMQNj7ROm7k0LyHaWE_w.png 666w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_Z4EMQNj7ROm7k0LyHaWE_w-300x162.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_Z4EMQNj7ROm7k0LyHaWE_w-610x330.png 610w\" sizes=\"(max-width: 666px) 100vw, 666px\" \/>\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-f6c7322 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"f6c7322\" 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-e7a92f8\" data-id=\"e7a92f8\" 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-cb425ce elementor-widget elementor-widget-text-editor\" data-id=\"cb425ce\" 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>These data can be fitted similarly well (e.g., in terms of\u00a0<a href=\"https:\/\/en.wikipedia.org\/wiki\/Root-mean-square_deviation\" target=\"_blank\" rel=\"noreferrer noopener\">RMSE<\/a>) by several structurally different models. For example:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-9aafb0f elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"9aafb0f\" 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-2df7843\" data-id=\"2df7843\" 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-a309451 elementor-widget elementor-widget-image\" data-id=\"a309451\" 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 loading=\"lazy\" decoding=\"async\" width=\"666\" height=\"360\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_7yJeZRlfYLsYaKwkOAsbuA.png\" class=\"attachment-large size-large wp-image-33275\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_7yJeZRlfYLsYaKwkOAsbuA.png 666w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_7yJeZRlfYLsYaKwkOAsbuA-300x162.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_7yJeZRlfYLsYaKwkOAsbuA-610x330.png 610w\" sizes=\"(max-width: 666px) 100vw, 666px\" \/>\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-827b38d elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"827b38d\" 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-3a60388\" data-id=\"3a60388\" 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-60ae162 elementor-widget elementor-widget-text-editor\" data-id=\"60ae162\" 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>The three models and their estimated parameters are as follows:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-903b803 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"903b803\" 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-9e92013\" data-id=\"9e92013\" 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-5dc44df elementor-widget elementor-widget-image\" data-id=\"5dc44df\" 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 loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"201\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw-1024x201.png\" class=\"attachment-large size-large wp-image-33276\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw-1024x201.png 1024w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw-300x59.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw-768x151.png 768w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw-610x120.png 610w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw-750x147.png 750w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_M6rcNtkd-A3pAUtJoibTNw.png 1121w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/>\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-05bb1e2 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"05bb1e2\" 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-cef88ac\" data-id=\"cef88ac\" 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-a5420c7 elementor-widget elementor-widget-text-editor\" data-id=\"a5420c7\" 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>As we can see, Model 1 is a 2nd-degree polynomial, Model 2 \u2014 3rd-degree polynomial, and Model 3 \u2014 exponential decay process. Obviously, these models imply different underlying mechanisms of how\u00a0<em>y<\/em>\u00a0changes over time.<\/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-553ac56 elementor-widget elementor-widget-text-editor\" data-id=\"553ac56\" 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>In fact, this toy example is based on data simulated from the following model that defines an exponential decay process:<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-2b0707c elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"2b0707c\" 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-c42673d\" data-id=\"c42673d\" 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-292338b elementor-widget elementor-widget-image\" data-id=\"292338b\" 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 loading=\"lazy\" decoding=\"async\" width=\"995\" height=\"191\" src=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_pUGwfYyJQ_wyIxUZrhRT1w.png\" class=\"attachment-large size-large wp-image-33277\" alt=\"\" srcset=\"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_pUGwfYyJQ_wyIxUZrhRT1w.png 995w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_pUGwfYyJQ_wyIxUZrhRT1w-300x58.png 300w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_pUGwfYyJQ_wyIxUZrhRT1w-768x147.png 768w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_pUGwfYyJQ_wyIxUZrhRT1w-610x117.png 610w, https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2020\/11\/1_pUGwfYyJQ_wyIxUZrhRT1w-750x144.png 750w\" sizes=\"(max-width: 995px) 100vw, 995px\" \/>\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-22d5c7f elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"22d5c7f\" 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-16e4679\" data-id=\"16e4679\" 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-9a5bd7e elementor-widget elementor-widget-text-editor\" data-id=\"9a5bd7e\" 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>Thus, Model 3 corresponds to the actual process that generated the data, and it fits that process quite well (in terms of the parameter estimates). However, we would never know this in real life and might wrongly conclude that some other model provides a better description of the underlying process. In practice, using the wrong model for decision making could lead to unforeseeable negative consequences (Grimm et al. 2020).<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-4e51d86 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"4e51d86\" 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-188449d\" data-id=\"188449d\" 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-5553001 elementor-widget elementor-widget-heading\" data-id=\"5553001\" 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\">Conclusion<\/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=\"has_eae_slider elementor-section elementor-top-section elementor-element elementor-element-72a1f80 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"72a1f80\" 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-30d9107\" data-id=\"30d9107\" 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-30cbcf2 elementor-widget elementor-widget-text-editor\" data-id=\"30cbcf2\" 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>Business users of Machine Learning models often ask to \u201cexplain\u201d how models make their predictions. Unfortunately, providing such an explanation is not always possible. This is especially true for complex models whose main purpose is to make accurate predictions. Arguably, most of the models built by Data Scientists nowadays belong to this very category. Nevertheless, engaging stakeholders early on in the project and demonstrating that the model captures their domain knowledge and\u00a0<a href=\"https:\/\/hackernoon.com\/explainable-ai-wont-deliver-here-s-why-6738f54216be\" target=\"_blank\" rel=\"noreferrer noopener\">is well tested<\/a>\u00a0can help in building their trust toward that 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-8e5de6d elementor-widget elementor-widget-text-editor\" data-id=\"8e5de6d\" 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>This article was originally published on <a href=\"https:\/\/towardsdatascience.com\/so-your-stakeholders-want-an-interpretable-machine-learning-model-6b13928892de\" target=\"_blank\" rel=\"noreferrer noopener\" class=\"broken_link\">Medium<\/a>, and republished with the permission of the author.<\/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>Business users of Machine Learning models often ask to \u201cexplain\u201d how models make their predictions. Unfortunately, providing such an explanation is not always possible. This is especially true for complex models whose main purpose is to make accurate predictions. <\/p>\n","protected":false},"author":976,"featured_media":11210,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[183],"tags":[990,991,992,993],"ppma_author":[3672],"class_list":["post-14130","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-ml","tag-business-users","tag-machine-learning-model","tag-predictions","tag-stakeholders"],"authors":[{"term_id":3672,"user_id":976,"is_guest":0,"slug":"sergey-mastitsky","display_name":"Sergey Mastitsky","avatar_url":"https:\/\/www.experfy.com\/blog\/wp-content\/uploads\/2021\/05\/Sergey-Ivanov-1-150x150.jpeg","user_url":"http:\/\/www.aviva.com","last_name":"Mastitsky","first_name":"Sergey","job_title":"","description":"Sergey Mastitsky is Data Science Lead at Aviva. He is the author of Statistical Analysis and Visualisation of Data Using R, Data Visualisation Using ggplot2, Implementing Classification, Regression, and other Algorithms of Data Mining Using R, and Times Series Analysis Using R, All these books were written in Russian. He provides data science consulting services at <a href=\"http:\/\/nextgamesolutions.com\/#slide=1\" target=\"_blank\" rel=\"noopener\">Next Game Solutions<\/a>"}],"_links":{"self":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/14130","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\/976"}],"replies":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/comments?post=14130"}],"version-history":[{"count":4,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/14130\/revisions"}],"predecessor-version":[{"id":33280,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/posts\/14130\/revisions\/33280"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/media\/11210"}],"wp:attachment":[{"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/media?parent=14130"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/categories?post=14130"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/tags?post=14130"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.experfy.com\/blog\/wp-json\/wp\/v2\/ppma_author?post=14130"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}