The various procedures are used in the literature for defining battery parameter change such as direct measurement methods, model-based methods, and data-driven methods, which contain the algorithms used in this paper also. The main aim of this study is to present a powerful and highly correct way of parameter forecasting of the A123 Systems 26650 cylindrical type Li-FePO4 battery cell. A few of the goal of this paper is to show the guessing performance of the artificial bee colony algorithm, which has a very limited number of applications on the battery parameter of literature, under the non-periodic dynamic charge/ discharge current profile. And then, a comparison has been made between artificial bee colony algorithm, artificial neural networks, support vector machine, and decision tree algorithm algorithms used in the paper. The load-connected terminal voltage is defined by considering the 100%-60% state of charge range in the primary usage areas of the batteries. A statistical comparison has been made by considering the maximum-minimum and average values of the real absolute errors, the maximum-minimum and average values of the squared errors, and R2 information regarding the results presented by the methods. Consequently, the regression values of the bee colony, neural network, support vector, and decision tree methods have been determined as 99.92%, 99.75%, 96.00% and 95.79%, respectively. Moreover, the mean value of the sum of squared errors of the methods has been calculated as 0.00202%, 0.00648%, 0.00998%, and 0.11%, respectively. As a new generation algorithm, artificial bee colony, which gave the most successful results according to the results obtained in the study, has been compared with two different methods selected from the existing literature, eXtreme Gradient Boosting and Smoothed eXtreme Gradient Boosting.