I am using XLSTAT tool to get the ML results. For the classification and regression trees, I am getting the results given in the image [https://i.stack.imgur.com/fa8mw.png]. I am not sure about the interpretation of the test statistic and p-value given in the table. Can I rely on them to say whether the prediction model is significant. For example, because p-value is less than 0.05, the model is statistically significant predictor? If yes, my problem would be, what about the other nodes in the model. Should I take my conclusion based on the p-value for the main node (node 1) only?

Regarding classification and regression random forest, XLSTS provides me with variable importance (mean decrease accuracy). Based on this number, I am not sure how to decide whether the dependent variable is a good predictor. For example, what if that number is negative, what if it is between 0 and 1, ..etc. Is there a rule of thumb?

  • $\begingroup$ Welcome to Cross Validated! These are really two questions, each of which warrants its own post. For the issue of an overall p-value, it will depend on what is being tested. Since you posted that question here, perhaps you would want to focus this post on the issue of variable importance. Then you can go back to the linked question from a few days ago and clarify what you want to do, as whuber and I asked in the comments. I would like to post an answer over there, but I want to be sure I am answering your question. Could you please clarify? $\endgroup$
    – Dave
    Mar 12 at 18:31


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.