Mirkan Emir Sancak

Mirkan Emir Sancak

This paper presents a reproducible comparison of cubic and radial basis function (RBF) interpolants for multivariate surface analysis. To eliminate evaluation bias, both methods are assessed under a unified slice-wise train/test protocol on the same synthetic function family. Performance is reported using RMSE, MAE, and $R^2$ in two regimes: (i) noise-free observations and (ii) noisy observations. In the noise-free regime, both interpolants achieve high accuracy with output-dependent advantages. In the noisy regime, exact interpolation overfits noisy nodes and degrades out-of-sample performance for both methods; in our experimental setting, the cubic interpolant is comparatively more stable. All experiments are fully reproducible through a single SciPy/NumPy-based script with a fixed random seed, repeated splits, and bootstrap-based uncertainty summaries. From an environmental engineering perspective, the main practical implication is that noisy or apparently inconsistent measurements in thermodynamic process systems should not be discarded by default; instead, they can be structured and interpolated to recover physically meaningful process behavior.

Mirkan Emir Sancak, Unal Sen, Ulker Diler Keris-Sen

Accurate determination of total oxidant concentration ([Ox]_{tot}) in non-thermal plasma (NTP)-treated aqueous systems remains a critical challenge due to the transient nature of reactive oxygen and nitrogen species and the subjectivity of conventional titration methods used for [Ox]_{tot} determination. This study introduces a novel, color-based computer analysis (CBCA) method that integrates advanced image processing with machine learning (ML) to quantify colorimetric shifts in potassium iodide (KI) solutions during oxidation. First, a custom-built visual data acquisition system captured high-resolution video of the color transitions in a KI solution during oxidation with an NTP system. The change in [Ox]_{tot} during the experiments was monitored with a standard titrimetric method. Second, the captured frames were processed using a robust image processing pipeline to extract RGB, HSV, and Lab color features. The extracted features were statistically evaluated, and the results revealed strong linear correlations with the measured [Ox]_{tot} values, particularly in the saturation (HSV), a and b (Lab), and blue (RGB) channels. Subsequently, the [Ox]_{tot} measurements and the extracted color features were used to train and validate five ML models. Among them, linear regression and gradient boosting models achieved the highest predictive accuracy (R^2 > 0.990). It was also found that reducing the feature set from nine to four resulted in comparable performance with improved prediction efficiency, especially for gradient boosting. Finally, comparison of the model predictions with real titration measurements revealed that the CBCA system successfully predicts the [Ox]_{tot} in KI solution with high accuracy (R^2 > 0.998) even with a reduced number of features.