{"id":4591,"date":"2022-11-10T17:48:04","date_gmt":"2022-11-11T01:48:04","guid":{"rendered":"https:\/\/live-cometml.pantheonsite.io\/?p=4591"},"modified":"2025-04-24T17:16:38","modified_gmt":"2025-04-24T17:16:38","slug":"model-interpretability-part-2-global-model-agnostic-methods","status":"publish","type":"post","link":"https:\/\/www.comet.com\/site\/blog\/model-interpretability-part-2-global-model-agnostic-methods\/","title":{"rendered":"Model Interpretability Part 2: Global Model Agnostic Methods"},"content":{"rendered":"\n
\"\"\/<\/figure>\n\n\n\n

Photo by NASA<\/a> on Unsplash<\/a><\/p>\n\n\n\n

\n

As mentioned in Part 1 of Model Interpretability<\/a>, the flexibility of model-agnostics is the greatest advantage, being the reason why they are so popular. Data Scientists and Machine Learning Engineers can use any machine learning model they wish as the interpretation method can be applied to it. This allows for the evaluation of the task, and the comparison of the model interpretability much simpler.<\/p>\n

Part 2 of this series about Model Interpretability is about Global Model Agnostic Methods. To recap:<\/p>\n

Global Interpretability<\/strong> aims to capture the entire model. It focuses on the explanation and understanding of why the model makes particular decisions, based on the dependent and independent variables.<\/p>\n

Global Methods<\/h1>\n

Global methods are used to describe the average behavior of a machine learning model, making them of great value when the engineer of the model wants a better understanding of the general concepts of the model, its data, and how to possibly debug it.<\/p>\n

I will be going through three different types of Global Model Agnostic Methods.<\/p>\n

Partial Dependence Plot (PDP)<\/h1>\n

The Partial Dependence Plot<\/strong> shows the functional relationship between the set of input features and how it affects the prediction\/target response. It explores how the predictions are more dependent on specific values of the input variable of interest over others.<\/p>\n

It can show if the relationship between the target response and a feature is either linear, monotonic, or more complex. It helps researchers and data scientists\/engineers understand and determine what happens to model predictions as various features are adjusted.<\/p>\n

According to Greenwell et al\u2019s paper: A Simple and Effective Model-Based Variable Importance Measure<\/a>, a flat partial dependence plot indicates that the feature is not important and has no effect on the target response. The more the Partial Dependence Plot varies, the more the feature is important to its prediction.<\/p>\n

When using numerical features, the importance of these features can be defined as the deviation of each unique feature value from the average curve, using this formula:<\/p>\n

\n
\n
\"\"<\/figure>
<\/picture><\/div>\n
<\/div>\n<\/div>\n<\/figure>\n

An example:<\/h2>\n

Let\u2019s say we are using the cervical cancer dataset<\/a> which explores and indicates the risk factors of whether a woman will get cervical cancer.<\/p>\n

In this example, we fit a random forest to predict whether a woman might get cervical cancer based on risk factors such as the number of pregnancies, use of hormonal contraceptives, and more. We use a Partial Dependence Plot to compute and visualize the probability of getting cancer based on the different features.<\/p>\n

\n
\n
\"\"<\/figure>
Source: <\/picture>christophm<\/a><\/div>\n<\/div>\n<\/figure>\n

Above are two visualizations that show the Partial Dependence Plots of cancer probability based on the features: age and years of hormonal contraceptive use.<\/p>\n

For the age feature, we can see that the PDP remains low until the age of 40 is reached, then the probability of cancer increases. This is the same for the contraceptive feature, after 10 years of using hormonal contraceptives, there is an increase in the probability of cancer.<\/p>\n

Advantages:<\/strong><\/p>\n