seaskythe / machine-learning-recommender-systems

This work was developed as a research for São Paulo State University (UNESP). Recommendation Systems are applications that have become very fundamental with the massive growth of the Internet, as they started to help the customers of different services to choose products that fit the most into their interest in an automated way. Additionally, with the increase in data volume, it is essential to use computational approaches for the automatic recommendation task. The implementation of these systems has been, therefore, crucial to the efficient operation of a lot of different businesses, being within the main ones, the streaming services. With the increasing popularity of these services, which have so many options, whether they are products for purchase or content for immediate consumption (such as media as TV Series), the need arose to increasingly improve the techniques used in the recommendation systems applied within these services, as the large number of options sometimes are not attractive to the costumer, being more efficient to show items that are into the area of interest of this customer. We used some data manipulation techniques in a movie dataset, to be able to create machine learning models that are capable of generating recommendations, graphically checking their effectiveness and performance in generating such recommendations.

Machine-Learning-Recommender-Systems

This work was developed as a research for São Paulo State University (UNESP). Recommendation Systems are applications that have become very fundamental with the massive growth of the Internet, as they started to help the customers of different services to choose products that fit the most into their interest in an automated way. Additionally, with the increase in data volume, it is essential to use computational approaches for the automatic recommendation task. The implementation of these systems has been, therefore, crucial to the efficient operation of a lot of different businesses, being within the main ones, the streaming services. With the increasing popularity of these services, which have so many options, whether they are products for purchase or content for immediate consumption (such as media as TV Series), the need arose to increasingly improve the techniques used in the recommendation systems applied within these services, as the large number of options sometimes are not attractive to the costumer, being more efficient to show items that are into the area of interest of this customer. We used some data manipulation techniques in a movie dataset, to be able to create machine learning models that are capable of generating recommendations, graphically checking their effectiveness and performance in generating such recommendations.

To use the application, the MovieLens 25m dataset is needed. You can download it in the following link: https://files.grouplens.org/datasets/movielens/ml-25m.zip After the download just put it in the "dataset" folder, and execute the notebooks (just remember to execute the "Prototipo I" before any other, because it has the data processing needed for the work).