Lusia Rakhmawati
Universitas Negeri Surabaya, Indonesia

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Comparative Evaluation of Moving Average Filtering and Linear ‎Regression Calibration in an ESP32-Based Water pH Sensor Mukhammad Indra Ukasya; Farid Baskoro; Rifqi Firmansyah; Lilik Anifah; Lusia Rakhmawati
ISEJ : Indonesian Science Education Journal Vol. 7 No. 3 (2026): June-September
Publisher : Yayasan Darussalam Bengkulu

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62159/isej.v7i3.2193

Abstract

This study compared Moving Average filtering and Linear Regression calibration in an ESP32-based low-cost pH measurement system using nominal buffers of pH 4.01, 6.86, and 9.18. One ESP32–PH-4502C sensor unit was evaluated through 60 repeated measurement cycles per buffer under approximately 25 °C room conditions. Performance was assessed using MAE, MSE, RMSE, bias, standard deviation, 95% confidence interval, drift slope, and lag-1 autocorrelation. Moving Average filtering did not reduce measurement error, and MAE generally increased with larger window sizes. Linear Regression reduced MAE by 25.2% at pH 4.01 but increased MAE by 158.9% at pH 6.86 and 42.4% at pH 9.18. The recorded Linear Regression condition also showed lower dispersion, drift slopes closer to zero, and lower lag-1 autocorrelation. However, because the processing conditions were not evaluated as paired transformations of the same underlying voltage sequence, these stability differences cannot be attributed solely to the regression transformation. The findings show that smoothing did not improve accuracy, while regression performance varied across buffer values. The results are device-specific and do not establish field performance or generalization across sensor units.