You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, i'm using the universal recommander template with predictionio to recommend products from a store. My issue is that said store can delete products in back office, for any reason, when it does the item must of course not be recommanded anymore. I then need a way to delete said item but not in a permanent way. Right now the best way i'v found is to give an item a specific property (let's say "eliminated"), re-train and then filter for on that property with a bias = 0, this way i just need to remove that property if i want the item back .
Is there a better way to do this? Re-train take time , does a way to avoid it exist?
I tried some other methods like an $unset event on the item , but doing so still require a re-train and it also require a full $set event to bring it back if i need to do so , and $delete event, that seems to do nothing at all, but it worked when i tried with a different template (similar product) and without a re-train. last thing i tried was the avaible/expire date but first i don't know the expire date until someone decide to actually make it expire ,second i need to make the change and then re-train if i want the changes to apply, so i'm back to square one.
Any answer will be very appreciated.
universal recommander version is 0.7.3
predictionio version is 0.13.0
The text was updated successfully, but these errors were encountered:
Hi, i'm using the universal recommander template with predictionio to recommend products from a store. My issue is that said store can delete products in back office, for any reason, when it does the item must of course not be recommanded anymore. I then need a way to delete said item but not in a permanent way. Right now the best way i'v found is to give an item a specific property (let's say "eliminated"), re-train and then filter for on that property with a bias = 0, this way i just need to remove that property if i want the item back .
Is there a better way to do this? Re-train take time , does a way to avoid it exist?
I tried some other methods like an $unset event on the item , but doing so still require a re-train and it also require a full $set event to bring it back if i need to do so , and $delete event, that seems to do nothing at all, but it worked when i tried with a different template (similar product) and without a re-train. last thing i tried was the avaible/expire date but first i don't know the expire date until someone decide to actually make it expire ,second i need to make the change and then re-train if i want the changes to apply, so i'm back to square one.
Any answer will be very appreciated.
universal recommander version is 0.7.3
predictionio version is 0.13.0
The text was updated successfully, but these errors were encountered: