INFUSE: Interactive Feature Selection for Predictive Modeling of High Dimensional Data

By Josua Krause, Adam Perer, Enrico Bertini. 2014

183 citations recorded by the source corpus.

Predictive modeling techniques are increasingly being used by data scientists to understand the probability of predicted outcomes. However, for data that is high-dimensional, a critical step in predictive modeling is determining which features should be included in the models. Feature selection algorithms are often used to remove non-informative features from models. However, there are many different classes of feature selection algorithms. Deciding which one to use is problematic as the algorithmic output is often not amenable to user interpretation. This limits the ability for users to utilize their domain expertise during the modeling process. To improve on this limitation, we developed INFUSE, a novel visual analytics system designed to help analysts understand how predictive features are being ranked across feature selection algorithms, cross-validation folds, and classifiers. We demonstrate how our system can lead to important insights in a case study involving clinical researchers predicting patient outcomes from electronic medical records.

Subfields: Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

In Signal Processing

Citations in the corpus

Recent advances and emerging challenges of feature selection in the context of big data (2015, methodology): “[159] are an interesting line of research.”
Visual exploration of machine learning results using data cube analysis (2016, methodology): “Researchers have utilized features to explain how they affect models [12, 11, 10, 5]: Kulesza et al.”
Interpreting Black-Box Classifiers Using Instance-Level Visual Explanations (2017, background): “A more recent solution is the idea of generating local explanations that are able to explain how a model makes a decision for single instances of a data set [11, 13, 20].”
A Workflow for Visual Diagnostics of Binary Classifiers using Instance-Level Explanations (2017, background): “In previous work, local explanations have been used to diagnose how models make decisions for single instances of a data set [15, 17, 27].”
ActiVis: Visual Exploration of Industry-Scale Deep Neural Network Models (2017, methodology): “Researchers have utilized features to visually describe how the models captured the structure of datasets [8, 20, 21, 23].”
A Multidisciplinary Survey and Framework for Design and Evaluation of Explainable AI Systems (2018, methodology): “tions in XAI studies. Satisfaction Measures Evaluation Methods User Satisfaction Interview and Self-report ([25, 62, 124, 125]) Likert-scale Questionnaire ([39, 62, 112, 124, 125]) Expert Case Study ([95, 106, 128, 130, 193]) Explanation Usefulness Engagement with Explanations ([39]) Task Duration a”
Gamut: A Design Probe to Understand How Data Scientists Understand Machine Learning Models (2019, methodology): “Example tasks include interactive model debugging and performance analysis [2, 44, 53], feature ideation and selection [6, 34], instance subset inspection and comparison [30, 31], model comparison [67], and constructing interpretable and trustworthy visual analysis systems [11].”