Sales and Subscriber prediction for Hubble Connected

The Requirement:

Hubble Connected wants to know what their sales and subscribers would be in the next few months to start campaingns, discounts, etc, if their expectations are not met.

How did I solve it?

• Used Multivariate Polynomial Regression to forecast sales from previous year data
• Used Long-Short-Term-Memory (LSTM) architecture to account for the periodic nature of subscriptions using sales, history and free trials data

Technologies:

• Programming Language: Python
• ML Models: Scikit-Learn, PyTorch

Team:

• Abhimanyu Bellam