Understanding the effect of uncertainty and noise in data on machine learning models (MLM) is crucial in developing trust and measuring performance. In this paper, a new model is proposed to quantify uncertainties and noise in data on MLMs. Using the concept of signal-to-noise ratio (SNR), a new metric called deterministic-non-deterministic ratio (DDR) is proposed to formulate performance of a model. Using synthetic data in experiments, we show how accuracy can change with DDR and how we can use DDR-accuracy curves to determine performance of a model.
@article{arxiv.2412.05882,
title = {Towards Modeling Data Quality and Machine Learning Model Performance},
author = {Usman Anjum and Chris Trentman and Elrod Caden and Justin Zhan},
journal= {arXiv preprint arXiv:2412.05882},
year = {2024}
}