Feature construction using explanations of individual predictions

Abstract

Feature construction can contribute to comprehensibility and performance of machine learning models. Unfortunately, it usually requires exhaustive search in the attribute space or time-consuming human involvement to generate meaningful features. We propose a novel heuristic approach for reducing the search space based on aggregation of instance-based explanations of predictive models. The proposed Explainable Feature Construction (EFC) methodology identifies groups of co-occurring attributes exposed by popular explanation methods, such as IME and SHAP. We empirically show that reducing the search to these groups significantly reduces the time of feature construction using logical, relational, Cartesian, numerical, and threshold num-of-N and X-of-N constructive operators. An analysis on 10 transparent synthetic datasets shows that EFC effectively identifies informative groups of attributes and constructs relevant features. Using 30 real-world classification datasets, we show significant improvements in classification accuracy for several classifiers and demonstrate the feasibility of the proposed feature construction even for large datasets. Finally, EFC generated interpretable features on a real-world problem from the financial industry, which were confirmed by a domain expert.

arXiv comment54 pages, 10 figures, 22 tables

References in corpus (13)

• XGBoost: A Scalable Tree Boosting System
• A Survey on Explainable Artificial Intelligence (XAI): Towards Medical XAI
• Learning Important Features Through Propagating Activation Differences
• AutoML: A Survey of the State-of-the-Art
• "What is Relevant in a Text Document?": An Interpretable Machine Learning Approach
• Consistent feature attribution for tree ensembles
• RUDDER: Return Decomposition for Delayed Rewards
• One button machine for automating feature engineering in relational databases
• Benchmarking and Survey of Explanation Methods for Black Box Models
• Quantifying and Visualizing Attribute Interactions
• Interpreting Classifiers through Attribute Interactions in Datasets
• Global Explanations of Neural Networks: Mapping the Landscape of Predictions
• Neural Feature Learning From Relational Database