Determination of Soluble Metals in Water using RGB Component Analysis and Machine Learning Models

Dissolved metals in water are one of the important parameters to be measured before it can be used for various purposes. This study aimed to develop a machine learning model to analyze tannic color reaction to soluble metals in water. Specifically, this also aims to determine the accuracy of the developed system and optimize the machine learning models through feature selection and dimensionality reduction. Lastly, the developed models were compared to the result of Atomic spectophotometry. To achieve these, development of models was done. The data set used was the tannic color reaction chart to soluble metals including the time elapsed for a color reaction to happen. The RGB components of each color was identified using the color segmentation feature of MatLab Software. In identifying the soluble metals, support vector machine (SVM), k-nearest neighbor (KNN) and decision tree was used while for determining the amount of soluble metal, the Multiple Linear and Support Vector Regression models were developed. In identification of soluble metals, results reveal that maximum accuracy, and performance was achieved using the optimized features for k-10 fold validation. In determining the dissolved amount, support vector regression outperformed the multiple regression model in terms of r2 values. Lastly, there is no statistical difference between the results produced by the model and the standard procedures in determination of soluble metals in water.