I’m learning regression and trying regression datasets.
I built following simple model
Now I need to estimate the future performance of algorithm (Question 1). How do I do that in Knime.?
Also what would be performance expectation (Question 2)? How can I be sure that the estimation is close to true performance(Question 3).?
Question 1:you can use the scorer node on the data you reserved on the second port to test.
Question 2: The statistics you find as output on the scorer node answers this question.
Question 3: If the data you used to train/test is a random sample of the data you’ll be using your model on in general it should perform similarly. Analyze the results and try to be sure you are not overfitting
I don’t quite get your point. This is what I get after running the workflow.
How can I tell the future performance ? by using R^2 ? Did I understand correctly that R^2 is the answer ?
Hi. The different statistics should help you.
R² is a measure of the degree of fitness of your model to your data. It measures the percentage of the variation of y explained by your model.
The mean absolute error is the average of the absolute value of the differences between real and forecasted values.
Chapter 3 of https://www.statlearning.com/ explains the theory behind and should help you to interpret the output of the node
Sorry for asking more. I don’t have access to this book. Can you please explain in simple terms how can I answer all these by using my simple model?
HI. The book is free to download in the link sent . If you check 3.1.3 you’ll have the basis to answer your question.
There is no need to use Normalizer Apply after the Normalizer node. The apply is for your test set.
Sorry. Can you elaborate it a little bit? Thanks
Thanks dude. Let me read. I will post my findings.
Here’s a recent blog post from our Evangelism team covering much the same ground:
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.