5/21

• researched different types of recommendation systems - 1 hour

5/22

• played around code from online that recommends different stuff - 1 hour

5/24

• tried to use matrix factorization to solve the problem of too little likes - 1 hour

5/25

• tried to use cosine similarity to get similar songs based on songs - 1 hour

5/26

• read a lot of articles on content based filtering and how to use user and song features for recommendations - 2 hours

5/28

• read multiple articles with examples that uses cosine similarity - 1 hour
• battling against the memory error when trying to use cosine similarity in python- 1 hour
• finished up cosine similarity as a recommendation system that recommends n songs from an input of songs - 3 hours

5/31

• looked for datasets that could be used to connect user features to songs since we don't have enough users and data ourselves - 15 minutes
• downloaded and analyzed the dataset that I found - 15 minutes
• data processing and training an ANN to predict user ratings for certain genres - 5 hours
• connected this recommendation system that uses user features to the one that uses song features - 2 hour

6/6

• blog on my content based recommendation system - 2 hour

Total Hours: 20 hours 30 minutes