Explaining models built in H2O
Evaluating single models based on global and local explanations
Machine Learning explainability refers to the ability to understand and interpret the decisions and predictions made by a machine learning model. Explainability is crucial for ensuring the trustworthiness and transparency of machine learning models, particularly in high-stakes situations where the consequences of incorrect predictions can be significant.
Today, several techniques are available to improve the explainability of a machine learning model, such as visualizing the model's decision-making process and using techniques like feature importance analysis to understand the factors that most influence the model's predictions. Some algorithms, like generalized linear models, decision trees, and generalized additive models, are designed to be inherently interpretable. Other algorithms like Gradient Boosting trees and neural networks are infamous for their black-box nature. For them, Post Hoc techniques provide a way to explain the predictions or decisions made by a machine learning model after the model has been trained.
H2O-3 is a fully open-source, distributed, in-memory machine-learning platform with linear scalability. Apart from supporting the most widely used statistical & machine learning algorithms, it also has a dedicated…
