Forecasting problem is a challenging problem on which many researchers from both field and academia work. This problem can also be considered as predicting the future. One of the most effective tools to predict the future is to utilize time series analysis. In real world, time series has uncertainty due to various situations. Conventional methods could not be sufficient to analyze such real world time series. Fuzzy time series methods have been proposed to analyze such time series. Many forecasting applications have been successfully performed by using various fuzzy time series forecasting models in the last decades. In fuzzy time series approach, determining fuzzy logic relationships between observation is a crucial process to reach accurate forecasting results. Determining fuzzy relations process is also called as fuzzy inference stage. In this stage, learning from the data is carried out by utilizing a method. Using a machine learning algorithm to accomplish this learning task would be a wise strategy. Therefore, there have been some fuzzy time series studies including machine learning algorithms in the literature. Artificial neural networks method is a core method in machine learning and this method has proved its success in many real world applications. In this study, it is explained how artificial neural networks method can be utilized to increase the forecasting performance of fuzzy time series approach. For the aim of this purpose, one of the basic fuzzy time series methods using artificial neural networks is examined. The algorithm of this fuzzy time series forecasting method based on machine learning is introduced. Also, by applying this method to a very well-known data, it is explained how this method works in details.