Description
Hello Dmitriy ,
First of all thank you for your work. I've been using Rsparse since a couple of weeks and everything is running smooth and well. The addition of the evaluation metrics is really a plus and I've been toying around with data to see which algorithm works best with the data I have (currently Linear Flow).
There is something though that I can't make properly work. The argument "items_exclude" on the predict function is not working as it should when we provide the article name (stringà). I provide the names (string) of the object to exlclude but they end up in the prediction. I've looked in the code, this seems to come from these lines:
items_exclude = match(items_exclude, private$item_ids)
item_embeddings = item_embeddings[, -items_exclude, drop = FALSE]I should have an output of 46 columns, but it gives me 122. I then use instead these lines which gives me the correct n of items:
items_exclude = Reduce(intersect, list(items_exclude,private$item_ids))
item_embeddings = item_embeddings[, !(colnames(item_embeddings) %in% items_exclude), drop=FALSE]When I provide the itemsID, it gives me the right number of items to recommend, unfortunately it is not the right products so I'll look into that now.