p********0 发帖数: 195 | 1 After I run the regression analysis by using Minitab, I find out there are
two predictors have significant relationship with the response, but the
other two don't. So I use the Best Subsets Regression to see what is the
best model ,and then find out the best model includes 3 predictors, which
means that it includes the predictor doesn't have significant relationship
with the response. What that means? This is the first time I use Minitab so
please help!!!!!!!!!!!!!!!!!!!!!!!!! |
p********0 发帖数: 195 | 2 waiting online.......... |
p********0 发帖数: 195 | 3 I have baozi as award!!! |
y**********l 发帖数: 19 | 4 The best subset in Minitab does forward selection. It start with a model
with one predictor, then two predictors,till all predictors. The Minitab
output give you a full list of those models. The minitab does not recommend
anyone from the list, you need to make decision based on the mellow-cp or r-
square (adj) value also shown in the Minitab output. I hope this can help
you. |
q**j 发帖数: 10612 | 5 the third variable might be relevant after you control for the first two.
what do you mean by saying "I find out there are
two predictors have significant relationship with the response, but the
other two don't." do you mean individually? |
d****z 发帖数: 301 | 6 这个正常吧,
forward或者backward一个一个地加/减才好吧.
不应该直接一次就删得就两个了的.
和minitab没有关系阿 |
p********0 发帖数: 195 | 7 Thank you very much. I just have a statistic class and now do some statistic
job in my internship.Thank you very much.
recommend
r-
【在 y**********l 的大作中提到】 : The best subset in Minitab does forward selection. It start with a model : with one predictor, then two predictors,till all predictors. The Minitab : output give you a full list of those models. The minitab does not recommend : anyone from the list, you need to make decision based on the mellow-cp or r- : square (adj) value also shown in the Minitab output. I hope this can help : you.
|
p********0 发帖数: 195 | 8 I compared the P-Value and alpha. But thank you.
【在 q**j 的大作中提到】 : the third variable might be relevant after you control for the first two. : what do you mean by saying "I find out there are : two predictors have significant relationship with the response, but the : other two don't." do you mean individually?
|