How can I rebuild the 100% Accuracy, from AutoModel, on the Iris Dataset with GLM or SVM? (RS 9.1)
geb_hart
New Altair Community Member
I only reach up to 96% and Im believing rapidminer does that via putting wheigts to the process? Or probably something different?
Thx for your Comments
Tagged:
0
Best Answer
-
Hi,
auto model got fixed seeds, to ensure the same results. I would recommend to change the seed in the AM process in the Split Data operator.
BR,
Martin5
Altair EmployeeAnswers
-
Hi,
Are you sure that this is not just because of a lucky split?
BR,
Martin0 -
I get the same results every time i try.. 100% for Generalized Linear Model, Deep Learning and SVMAutoModel on Iris Dataset, no adjustments ... (Am I the only one?
)0 -
Hi,
auto model got fixed seeds, to ensure the same results. I would recommend to change the seed in the AM process in the Split Data operator.
BR,
Martin5 -
Hi,Martin is correct. A random subset of the data is used for validation and if this subset is "easy" to be predicted, the accuracy can be a bit better than what you would get with a different data split or validation scheme.Best,Ingo1
