Decision Tree Based Wrappers for Hearing Loss

TL;DR

This paper presents FEDORA, an evolutionary feature engineering wrapper using decision tree models to optimize data for hearing loss screening, reducing features while maintaining or improving accuracy.

Contribution

It introduces FEDORA, a novel evolutionary feature engineering framework that effectively reduces data dimensionality and enhances model performance in hearing loss detection.

Findings

FEDORA achieves a maximum balanced accuracy of 76.2% with 57 features.

It can generate a single-feature model with 72.8% accuracy.

FEDORA outperforms traditional feature selection methods.

Abstract

Audiology entities are using Machine Learning (ML) models to guide their screening towards people at risk. Feature Engineering (FE) focuses on optimizing data for ML models, with evolutionary methods being effective in feature selection and construction tasks. This work aims to benchmark an evolutionary FE wrapper, using models based on decision trees as proxies. The FEDORA framework is applied to a Hearing Loss (HL) dataset, being able to reduce data dimensionality and statistically maintain baseline performance. Compared to traditional methods, FEDORA demonstrates superior performance, with a maximum balanced accuracy of 76.2%, using 57 features. The framework also generated an individual that achieved 72.8% balanced accuracy using a single feature.