2
$\begingroup$

I am fairly new to statistics and would love to make my basics clear. I am confused about what data type should a simple/multiple regression model has?

Can you please answer my question with an explanation? Also, please recommend a good book on regression?

$\endgroup$

2 Answers 2

0
$\begingroup$

Regression is a family of methods, not a single one.

I think you are probably referring to ordinary least squares regression. If so, then the dependent variable (DV) must be continuous (or nearly so) and the independent variable(s) can be continuous or categorical.

But there are also methods for dealing with categorical DVs (various kinds of logistic regression), time DVs with censoring (Cox proportional hazards and other survival methods) and others.

$\endgroup$
2
  • 1
    $\begingroup$ Even a (0, 1) dependent variable (I'd rather say response or outcome) can feature in a regression. That regression is sometimes called a linear probability model and has many able defenders, who regard any consensus that logit or probit makes more sense as just another point of view. I'd be perfectly happy with e.g. proportion female or proportion recovered from illness as an (approximate) continuous response, and that is just the same kind of data averaged and not fundamentally different. $\endgroup$
    – Nick Cox
    Commented May 5, 2020 at 12:44
  • 1
    $\begingroup$ Thanks, Peter!! $\endgroup$ Commented May 6, 2020 at 13:04
-2
$\begingroup$

For regression following variable types hold:

Dependent variable must be continuous only.

Independent variable may be continuous or discrete.

If dependent variable is discrete then the problem will become classification problem rather than regression.

As per me

An Introduction to Statistical Learning

is good book to learn regression.

$\endgroup$
0

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.