Recommender systems · Module 5

Interpreting
the Results.

Why did the model pick this suggestion rather than another? Learn to open the black box with SHAP, LIME and the essential performance metrics.

SHAP

Shapley values
+ LIME

92^%

Prediction
confidence

XAI

Explainable
AI methods

— Activity

Explain the recommendation

Interactive

— Use case

Marie rated Inception (5⭐), Interstellar (4⭐), Dune: Part Two (4.5⭐). The model strongly recommends Interstellar to a similar new user. Why?

The model (SVD + user-based collaborative filtering) identified that Marie belongs to a cluster of users who love philosophical science fiction / space exploration films with a strong emotional component.

— Key contributing factors

High cosine similarity with users who liked Inception + 2001: A Space Odyssey
Latent factor no. 3 strongly activated (themes: time, relativity, family)
SHAP contribution: +0.38 for "Christopher Nolan" and +0.29 for "space exploration"

Overall recommendation confidence: 92% probability that Marie gives ≥ 4/5

Precision@5

0.78

Relevant top 5

Recall@10

0.64

Coverage of good items

NDCG@10

0.81

Ranking quality

RMSE

0.86

Average prediction error

Computing & Generating

Home

Back to the home page

Interpretingthe Results.

Explain the recommendation

Why "Interstellar" for Marie?

Interpreting
the Results.