Open jonnylee719 opened 7 years ago
Not sure, exactly what score are you using?
Flavio
On Tue, Nov 8, 2016 at 11:21 AM, jonnyLee notifications@github.com wrote:
Hi, I have been trying out your implementation of fpmc and it's been doing great, so thank you! However, my question is on the scores computed using the learnt models. I computed the scores following the item recommendation method shown in the fpmc paper. The scores can be >1. I am uncertain if I should be expecting a score between 0 to 1. Are there supposed to be some bounds to the predicted scores or the score can be any real number such that an item ranking can be produced?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/flaviovdf/fpmc/issues/3, or mute the thread https://github.com/notifications/unsubscribe-auth/AAf08ICf78X7K8pN7CfZzcNdDgomBlguks5q8HdLgaJpZM4KsaoZ .
Hey thanks for the quick reply! I attach the screenshots to the mf and fmc formulas and the combined fpmc formula that I used to compute the scores of items.
Hi, I believe that there is no bound on these parameters. Check Figure 5 in the paper. Parameters are initialized following a Normal(0, 1)
Sorry it took a while to answer.
Flavio
On Tue, Nov 8, 2016 at 12:22 PM, jonnyLee notifications@github.com wrote:
Hey thanks for the quick reply! I attach the screenshots to the mf and fmc formulas and the combined fpmc formula that I used to compute the scores of items.
[image: mf] https://cloud.githubusercontent.com/assets/8478381/20102619/7fe545a4-a5a5-11e6-8791-e6b3e170a4c7.png
[image: fmc] https://cloud.githubusercontent.com/assets/8478381/20102618/7fe0ce52-a5a5-11e6-9b0e-4ddbabcef8cf.png
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/flaviovdf/fpmc/issues/3#issuecomment-259148767, or mute the thread https://github.com/notifications/unsubscribe-auth/AAf08LjEFcMuZbD9FSCipFcnNHtLb9-yks5q8IWigaJpZM4KsaoZ .
Hi, I have been trying out your implementation of fpmc and it's been doing great, so thank you! However, my question is on the scores computed using the learnt models. I computed the scores following the item recommendation method shown in the fpmc paper. The scores can be >1. I am uncertain if I should be expecting a score between 0 to 1. Are there supposed to be some bounds to the predicted scores or the score can be any real number such that an item ranking can be produced? Thanks for any help!