The Application Of Machine Learning In Recommendation System

As the advent of the Big Data era, several new modern subjects are becoming increasingly important and popular. Machine learning is one of those subjects, which has been widely and successfully applied in Informatics, Medical, Biology, Transportation and other fields. One of the most popular fields for applying machine learning is Finance and Economics, such as stocking price prediction and risk management and evaluation. In order to further learn how to apply machine learning into practice, it’s necessary to have a deep understanding of the state-of-art algorithms in machine learning, from their fundamental machinery to the conditions for which the algorithms are suit. Therefore, the major goal of this thesis is to strengthen my understanding of different machine learning techniques through using graphical representation and analyzing Yelp’s dataset to apply different machine learning algorithms to help Yelp build its own recommender system in which new users are wisely recommended based on the trained model. The algorithms include logistic regression, neural network and decision tree which includes simple decision tree, random forest and bagging tree. Then the performances of different models are evaluated by ROC curve. Furthermore, a ranked candidate set would be generated for each new user who needs recommendations.