# Movie Recommendation Engines

*Business Case*
Movies are well loved by much of the world's population! Today you have joined the ranks of Netflix's data science team; your mission is to recommend the best movie to each customer.

*Machine Learning Method*
In this project, you will learn about Factorization Machines. This is a highly scalable algorithm that was developed by Steffen Rendel in 2010. It has the capacity to leverage extremely large datasets at the Terabyte scale, while still training in linear time. 

![alt text](Images/recommender_1.png ) 

In essence, the factorization machines model is calculating the dot product between the user information and the item information, then computing the difference between those to update the model.

![alt text](Images/recommender_2.png ) 


In order to accomplish this, you'll need to *format your data as events*. Each row in your final data set will need to be a single point in time when a customer interacts with a product account. Each column will be either a binary indicator for the product/user, or another feature.

*Data set description*
We highly suggest using the Movie Lens dataset. This is well suited to recommendation problems and should be available from public sources.

*A follow up note on method*

* The built-in algorithm you are working with is designed to only handle either binary classification or regression. It can be extended to provide multi-class classification with KNN. First start with the recommender, and if your are up to the task, consider extracting the matrix and using it to build KNN clusters.

*Starter Code*
- https://github.com/aws-samples/amazon-sagemaker-architecting-for-ml/blob/master/Starter-Code/Recommendation-System-FM-KNN.ipynb

*Extension*
After you've run through a few steps, try extending your solution by using the managed recommendation service from AWS, Personalize. 
- https://github.com/aws-samples/amazon-personalize-samples 

*References*

* In Python: https://www.datacamp.com/community/tutorials/recommender-systems-python
* Intuitively: https://www.analyticsvidhya.com/blog/2018/01/factorization-machines/ 
* Formally: https://www.csie.ntu.edu.tw/~b97053/paper/Rendle2010FM.pdf