Data Science Recommendation Engine Fundamentals – Part 01

AI & ML & Blockchain, Data & Analytics, Machine Learning, Recommendation Engines
Product Cognizant > AI & ML & Blockchain > Machine Learning > Recommendation Engines > Data Science Recommendation Engine Fundamentals – Part 01

Data Science enthusiastic interested in recommendation engines (RE) mastery should first establish a fundamental understanding of RE.

Overall, a RE works around “Users -> Items -> Users & Items Interactions”

Below is a step oriented approach to building such fundamentals:

Step 1: Core elements to identify in your business under consideration
Step 2: Expand knowledge in your business systems on the above elements
Step 3: Usage of Recommendation Engines results in a business application

Step 1: Core elements to identify in your business under consideration

  1. User
  2. Item
  3. User – Items Interactions

Step 2: Expand knowledge in your business systems on the above elements

A. User

  1. Demographics
  2. Items Bookmarks or Wishlist
  3. Items Added to Cart or Removed (for real-time recommendations during a shopping session)
  4. All Items Purchased History
  5. Items Comments (includes experiences with images, if possible)
  6. Purchased Items Rankings or Ratings
  7. Liked or Unliked Items
  8. Environment based information including
  9. 1. Time spent on an item details page
  10. 2. Page navigation by the user (previous/next)
  11. 3. A page referrer (from where user landed into this a specific page of your web applications)

B. Items

  1. Items Catalog
  2. Item Descriptions (with indexing in search engines)
  3. Item Location Availability (for proximity recommendations)
  4. Item Value
  5. Items Related and weight of an item in an overall basket purchase

Step 3: Usage of Recommendation Engines results in a business application

  1. Homepage recommendations
  2. What next recommendations
  3. Recommendations through Email notifications (to bring back the user to your web application)
  4. Recommendations when the user is in an item page for related items or cross-selling

A pictorial representation of the above learnings:

Post Views: 489