Recomendation is one of the hottest topics in data science and machine learning. It can be used for recomending:

• movies on netflix;

• products on amazon; and

• songs on online music (Spotify or WSDM)

We wanted to explor this further.

Can we tell if a user will repeatedly listen to a specific song in the 1 month time frame after first hearing the song.

This can be used to play songs that the user is likely to enjoy, thus increasing their satisfaction with WSDM.

No insight into trends initially. How can we make predictions?

XGBoost

• Efficent
• Accurate
• Tree Based
• CART (Clasification And Regression Tree)

Accuracy: val-auc:0.767018

Expected: 0.5

Data Manipulation:

• Converting factor level data (e.g., user ID or song name) into numeric values.

• Scaling values with (Standardization)

\(x^{\prime}=\frac{x - \bar{x}}{\sigma}\)