In this Project we have trained different models on the hotel reviews data set, in order to predict review rate by different review features. We have started with geographic features combined with additional information on the reviewer and the hotel, but the resoult was not that good. Then, we have decided to focuse on semantic text analysis technices. This approach leads us to better resaults with nice correlation, after we have added the text features, but still the accuracy was not as we have expected. We put a lot of effort in the project and learned a lot from it. probably, we would have change some of the decision we made along the way and try other directions. The model with the best performence is MLPClassifier.