The purpose of the research is to apply a machine learning tool for predicting answers and to carry out an evaluation, which was done using manual accuracy evaluations. The theoretical part presents basic information about machine learning, data science and artificial intelligence. Machine learning is described in the context of information sciences, compared to related fields of expertise such as data mining and deep learning. A brief history, advantages and disadvantages, current applications and views on the future and development of machine learning are presented. A survey was conducted on 282 participants, where the answers were collected through a self-administered survey. In the test, two models were used, which differ from each other in the size of the data sets for training and testing in the ratio of 25/75 and 50/50. The models are intended to predict binary and numeric values and multiple selection values. The results of this research show that there is no unique forecasting model universally applicable to any database while achieving maximum efficiency. Model A showed an advantage in testing on sets of binary and numerical values, and Model B on sets of multiple choice answers. The choice of model depends largely on the type and size of the data set, and the choice is ultimately decided by the more prioritized data type.