1 Comment.
following up on Robyn Dawes's article, "The Robust Beauty of Improper Linear Models in Decision Making" - hugely important reading! - consider a recommender system:
- identify simple (input) criteria for each (output) decision
- assign simple weights (+1, 0, -1, whatever, but keep it simple)
- add up the weights selected for each decision
- sort into descending order and recommend the top few!
... a la SAnTA = "Structured Analytic Technique Advisor", the MITRE JavaScript browser-based "simplest recommender system that could possibly work" ... (^_^)
– z 2016-07-12 11:37 UTC</div>