Reply
Wed 9 Aug, 2017 12:51 am
Hi
It’s a little weird question. I was browsing for something else and by chance found an interesting page but cannot reach to that page or info again after trying really hard.
It was about determining significant data points (cut off) via linear regression for creating categorical variables. I will share an example; if I have two continuous variables, Age and Blood Sugar. I run linear regression and find that Age can predict Sugar values. Now, to determine particularly which age group would have higher Blood Sugar, I want to create a categorical variable “Age Group”. For example,
Age < 50 = 0
Age > 50 = 1
How can I determine the cut off points or ranges for Age Group to get the significant results via logistic regression? May be grouping the Age variable as <60 and > 60 may give a significant result rather than grouping < 50 and >50.
I hope I explained the question. Looking forward for your valuable input.
Thanks