CODEWITHSUNDEEP
orange
4thSpace
4thSpace

Reputation: 44310

What is discrete class variable in Orange?

I'd like to start trying to train stock price time series using Orange. I have a simple time series for Amazon that is attached to a Logistic Regression widget. The widget throws the error:

Discrete class variable expected.

Any one know what this means or how to solve it?

Workspace: https://www.dropbox.com/s/e43ssam3higoqgb/stockprice_regression.ows?dl=0

Data file: https://www.dropbox.com/s/38ye3qm92dpbiov/amazon.csv?dl=0

-- EDIT -- Logistics regression has been replaced with linear regression. This moves things along a little.

enter image description here

Upvotes: 0

Views: 4382

Answers (1)

Primoz
Primoz

Reputation: 1492

In order to learn the model in a supervised manner (Logistic regression is the model that need to be learned), you need to tell a model what is the class variable in your data. For Logistic regression which is a classification model, the class variable must be discrete (it represent few classes in data). Since you have a continuous class variable (a class variable is marked as a target in File widget) I suggest you use one of the regression models such as Linear regression (the model that do not predict classes but values directly).

Upvotes: 1

Related Questions