The buildup of scale in cooling systems, especially evaporative cooling systems, is frequently a significant problem because of calcium (Ca) ions in raw or makeup water. As water evaporates, the concentration of these ions increases, leading to the formation of insoluble salts such as calcium carbonate (CaCO₃). This scaling may decrease heat transfer, inefficiencies, and increased energy usage. The calcium must be removed for the cooling system to operate best. The present study investigated the removal of Ca²⁺ from cooling tower water using Amberlite IR120 and predictive machine learning approaches. A lab-scale ion exchange column was used in this study. Response surface methodology (RSM), artificial neural networks (ANNs), and adaptive neuro-fuzzy inference systems (ANFISs) were used to optimise and model calcium removal using Amberlite IR120. The effects of the following process parameters were studied: contact time (min), pH, concentration (mg/L), dosage (ml), and temperature (K). RSM was used for process optimisation. The ANN model construction used 70% of the data for training, 15% for testing, and 15% for validation. The network was trained using feed-forward propagation and the Levenberg–Marquardt algorithm. The ANFIS was generated using a grid partition and trained using a hybrid method; 80% was used for training, and 20% was used for checking. Regression (R²), mean square error (MSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and average relative error (ARE) were also used. Numerical optimisation yielded an optimal removal percentage of Ca²⁺ of 99.07% at 89.55 minutes, 4.17, 452.83 mg/L, 132.57 ml, and 295.58 K. The developed predictive machine learning model fits the three machine learning models with regressions of 0.9777, 0.9994, and 0.9903 for RSM, ANN, and ANFIS, respectively. This study has shown that machine learning is an effective tool for removing Ca²⁺ from cooling water Amberlite IR120 resins.