This study presents a new investigation on the improvement of robot learning for water analysis with the combination of decision making and machine learning (ML) processes for a robotic system. The aim of the study was to perform simulations for the distinction of drinking and undrinkable water for further implementation in a robot. The decision-making process was performed with the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). The data analysis with ML was done by using Microsoft Visual Studio. The TOPSIS analysis showed that the candidates having high values of profit criteria and low values of cost criteria had a better rank. The same result was obtained in the analysis of the physicochemical properties of water as well as its ingredients. The differences in the closeness coefficient values of the best and the worst candidates were 35%, and 45% in the first and second series of analyses, respectively. The ML simulation showed that using the modified code could improve the learning accuracy to 69%, which improved to 73% after using the Synthetic Minority Over-sampling Technique (SMOTE) for class balancing and applying GridSearchCV to tune the hyperparameters. The electronic components of a robotic system and the remote control of its prototype for further application of the current work have also been presented. The obtained results could be used for the implementation of the combined software in a robotic system for water analysis.