Next Recommendation Algorithm

[wfengk]

Hey guys,

Thinking the next iteration of the recommendation algorithm will use Azure Machine Learning (probably linear regression to start off with - we will see how well it works). It will be a REST endpoint that we send data to and will retrieve back a calculated value.

Here are my rough thoughts so far as to what should be considered in the algorithm:

---

Input (to Azure ML)

• Whether the price of restaurant matches user choice
• Does the restaurant match user's preferences
• Average rating of restaurant
• Average rating the user gives
• Review count of restaurant
• Is this restaurant one of the user's favourites?
• Optional - is the location too far away?
• Optional - has the user been to this restaurant before? (does the user like to try new places)

Output

• Predicted rating a user would give for a given restaurant

---

Our output recommendation would then be the highest predicted rating amongst a set of restaurants

Let me know what you think @RedTn @jrhender @gargirou