How False Data Affects Machine Learning Models in Electrochemistry

Aug 2, 2026

Uploaded by Maeve

Noisy and false input data can substantially degrade prediction quality in chemistry machine learning, yet model robustness varies across algorithms. This study examines failure input data effects in an electrochemistry setting using heteroatom-doped graphene supercapacitor data, training 12 standalone models plus a stacking ensemble while progressively adding noise. Error trends are analyzed via linear regression (MAE/MSE/RMSE/MAPE/R2), with SHAP and PDP visualizations to interpret how errors impact electrochemical features. Results compare noise sensitivity and base accuracy across linear, tree-based, SVM/KNN/NN, and stacking models.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages40

Reading Time1 h 41 min

AccessFree to download

Abstract

Noise and failure input data in electrochemistry

Model comparison and stacking approach

Error-metric trends and interpretability (SHAP, PDP)

Introduction

Data quality as a driver of ML research

Sources of noise and false measurements in experiments

Need for noise-robust, high-accuracy models

How False Data Affects Machine Learning Models in Electrochemistry? Krittapong Deshsorna ,b, Luckhana Lawtrakula, Pawin Iamprasertkuna, b, * aSchool of Bio-Chemical Engineering and Technology, Sirindhorn International Institute of Technology, Thammasat University, Pathum Thani, Thailand 12120 ([pawin@siit.tu.ac.th](pawin@siit.tu.ac.th)) bResearch Unit in Sustainable Electrochemical Intelligent, Thammasat University, Pathum Thani, Thailand 12120 KEYWORDS: Stacking Models, Ensemble, Machine Learning, Noise, Aleatoric, Epistemic TOC: Graphical Abstract ABSTRACT Noisy data is detrimental to the prediction of machine learning in chemistry. But some models are more tolerant to noise than others. The selection of machine learning models in electrochemistry is based on only the data distribution without concerning the noise of the data. This study aims to provide a discussion of the introduction of failure input data in electrochemistry, which demonstrated using heteroatom doped graphene supercapacitor data. The electrochemical data were tested with 12 standalone models including XGB, LGBM, RF, GB, ADA, NN, ELAS, LASS, RIDGE, SVM, KNN, DT, and our “stacking” model. By gradually adding the noise/false data into the pool, the models were then trained on both noisy and ground truth data to obtain various error metrics (MAE, MSE, RSME, MAPE, and R2) . The linear regression was then fitted on the increasing/decreasing errors to obtain the slope and intercept to discover estimated base accuracy (intercept) and noise sensitivity (slope) . Hence, this study utilized contour plots, SHAP, and PDP to explain how the error affects the electrochemical feature including prediction and analysis. It is found that linear models handle the false data well with an average MAE slope of 1.513 F g-1, but it suffers from prediction accuracy due to having an average MAE intercept of 60.20 F g-1. This is due to improper modelselection for this type of data (average R2 intercept of 0.25) . The “Tree-based” models fail in terms of noise handling (average MAE slope is 58.335 F g-1), but it can provide higher prediction accuracy (average MAE intercept of 30.03 F g-1) than that of linear models. Treebased models also fit well to the data (average R2 intercept of 0.9516) . This suggests that the linear based model can be well described the relationship between capacitance and surface area. While the “tree based” model can be used for handling the other electrochemical features e.g. amount of heteroatom doped, current density, and so on. Miscellaneous models such as SVM, KNN, and NN, are moderately robust to noise (average MAE slope of 25.956 F g-1) and provide moderate accuracy (average MAE intercept of 41.306 F g-1) . The models also fit moderately well to the data (average R2 intercept of 0.546) . To address the controversy between prediction accuracy and error handling, the “stacking model” was constructed, which not only shows high accuracy (MAE intercept of 24.29 F g-1), but it also exhibits good noise handling (MAE slope of 41.38 F g-1and R2 intercept of 0.86), making stacking models a relatively low risk and viable choice for electrochemist. This study presents that untuned NN is not suitable for electrochemical data, and improper tuning results in a model that is susceptible to noise, which directly affects the misleading in the electrochemical discussion. Thus, “STACK”models should provide better benefits in that even with untuned base models, it can achieve an accurate and noise tolerance. Overall, this work provides insight into machine learning modelselection for electrochemical data, which should aid the understanding of data science in chemistry and energy storage context. INTRODUCTION The “Fourth Industrial Revolution” signifies a profound shift in how we live, work, and conduct research. This new phase of human advancement, driven by remarkable technological progress akin to the first three industrial revolutions, marks a significa

FAQs

Why is false or noisy data harmful to machine learning in chemistry?

Noisy data can cause model predictions to deviate from the ground truth, leading to unreliable and inaccurate results.

How does the study evaluate the impact of false data on model performance?

It gradually adds noise/false data to the dataset, trains models on both noisy and ground-truth data, and computes error metrics such as MAE, MSE, RMSE, MAPE, and R2.

Which model families show different trade-offs between noise handling and prediction accuracy?

Linear models handle false data better but can lose accuracy, tree-based models show poor noise handling but higher accuracy, and SVM/KNN/NN are moderately robust. The stacking model aims to balance both by improving noise tolerance while maintaining accuracy.

Aug 2, 2026

Uploaded by Maeve