Reputation: 5791
I have training set and test set (csv files with header), in which I have to classify each value. There is 118.000 uniq values in X column, and only about 13000 in y1 column, so there will be 13000 categories.
From Training set I need only X
and y1
column to train model. I need to classify X value to one of categories (find normal from of initial word). I tried all multi algo but failed trying to evaluate model.
Visualizing Score model return this:
What can be a problem, it just returns -2 code as error and this log
UPD1: By Metadata Editor module under Project Column Module made column y1 as categorical, nothing seems to be changed
Upvotes: 0
Views: 91
Reputation: 5791
Moncef provided here the solution to my problem:
The point is that Azure has limitations on maximum categories 8192, this is why the number should be decreased by R or python modules or own evaluation module may be created. Or there is another way, evaluation step may be skipped, because model`ve been trained successfully.
Upvotes: 1