Interest in wireless sensors is on the rise due to the Internet of Things (IoT) action. This is because putting them in places that are not reachable by connected sensors is a distinct possibility. There has been a lot of progress in this area, but energy constraints are still holding things back. Therefore, optimizing sampling to a rate that captures relevant information without consuming unnecessary energy is crucial. An adaptive sampling algorithm called the dynamic sampling rate algorithm (DSRA) can be used to obtain sufficient sampling, but it needs a trained eye to adjust its parameters, which isn’t always accessible. The goal of this research is to refine this algorithm by making it more reliant on machine learning to adjust these settings. A redesigned algorithm and an optimization method taking into account a predefined error threshold was devised to accomplish this goal. Next, the method was tested with both real and simulated data, using a set of error criteria that had already been developed. The algorithm’s ability to efficiently gather data depending on the predefined error threshold was demonstrated by the findings, which also demonstrated adaptive sampling behavior. From the results, it can be reasonably concluded that the algorithm that was constructed allows sensors to adjust their sampling according to the signal’s amplitude.