Skip to content

Utility Matrix LLR

Jaime edited this page Aug 15, 2017 · 1 revision

Log Likelihood Ratio

The Log Likelihood Ratio (LLR) was popularized by Ted Dunning in his paper, “Accurate Methods for the Statistics of Surprise and Coincidence". The LLR relies on calculating similarity between two users or items based on statistics that revolve around occurrences related to these users or items.

The output of the module follows the format: ((USER-ID1, USER-ID2), LOG LIKELIHOOD RATIO)

((695858,2397231),0.78)
((2050389,654334),0.15)```
Clone this wiki locally