When to use Linear Regression and When to use Logistic regression - use cases

Linear Regression is used to find the relation between dependent variable and independent variable.

Logistic Regression is used when dependent variable is categorical.

But this doesn't exactly suffice, what use cases you can use either of these, or can we use a combination of them?

Logistic Regression is used when you know that the data is lineraly seperable/classifiable and the outcome is Binary or Dichotomous but it can extended when the dependent has more than 2 categories.

Linear Regression is used to find the relation and based on the relation between them you can predict the outcome, the dependent variable should be numeric.

What kind of usecases are you expecting? give an example so that we can extend it further.

• Linear Regression,

no class definition, the response variable is a continues value. Model the relationship between explanatory variables and the response variable. Suppose you want to estimate the house price based upon the area of the land, you can use linear regression. It is an equation of a straight line, where your response can take any value between -Infinity to + infinity.

• Logistic Regression,

Now suppose you want to fit a model for two class problem like ham or spam, or malign or benign tumour(ie response variable can only take 0 or 1). here, you got only two values for the response variable. Now, if you want to fit this with a linear model (which as we know can take any value between - inf to + inf) how would be interpret response value between 0 and 1 and how would you interpret values more than 1 an double than 1 and ten times of 1...(you are getting my point?) hence, in such cases, we tend to use Logit function, which can only take two values. In this case, we have to decide about the threshold for the decision boundary but once done, you would know your response values as 0 r 1.