I'll update this readme file

Recommendation systems play a crucial role in many services today and help involve and interact with their user base. Also, they help to connect people with the right products and industries. In this project, we aimed to make a recommendation system with the Neural Collaborative Filtering method. We used one of the most popular datasets, MovieLens Dataset, transformed the movie ratings as interactions, and fed them into our system. After creating the design and understanding the whole model, we started to see the effects of different parameters in the system. We analyzed the system in the case of 4 other parameters: Neuron Size, Number of Layers, Activation Function, Maximum Epoch, and Batch Size. We observed that increasing the epoch until a point affected the system positively. Then in the batch size, we found the optimal batch size as 512, while other sizes are also very close to optimal. Increasing the neuron size also affected the system positively until 64 neuron size. On the other hand, we couldn’t see a significant effect of layer size in the system. As a result, we got the optimal development of 0.86.

#collaborative filtering #matrix factorization #one-hot encoding