I am dealing with a classification problem and my target column consists of binary (0/1) values. And some of my independent columns are categorical and I give an example of one of them, for example there are 3 classes in the Class column (Business, Eco, Eco Plus). Now how can I check the relationship between this categorical value and binary target.
9
-
1$\begingroup$ Did you look at a contingency table? Then conditional probabilities? $\endgroup$– Ggjj11Commented Oct 2 at 12:10
-
$\begingroup$ You could use a logistic regression. If you have an ordinal independent variable, you may be interested in this other thread: stats.stackexchange.com/questions/101511/… or the many other ones on this subject: stats.stackexchange.com/search?q=logistic+ordinal+independent . @Ggjj11: OP mentions there are several independent variables (not just the "Class" variable), so a two-way contingency table may not be the best option here. $\endgroup$– J-J-JCommented Oct 2 at 12:21
-
$\begingroup$ In addition, your question may be lacking some details to give you a proper answer. Why do you want to check the relationship between your dependent variable and the categorical variables? It sounds like you know how to check the relationship between continuous variables and your dependent variables, but not categorical ones. Why is that? What have you done so far? Have you checked the relation between the dependent variables and the independent continuous ones? What kind of model have you been using for that? Do you encounter difficulties interpreting the output of this model? $\endgroup$– J-J-JCommented Oct 2 at 12:30
-
$\begingroup$ Related $\endgroup$– DaveCommented Oct 2 at 12:32
-
1$\begingroup$ I think you should upload part of your dataset so we can better understand your question. What is actually the question motivating the analysis ? $\endgroup$– CaroZCommented Oct 2 at 12:33
|
Show 4 more comments
1 Answer
$\begingroup$
$\endgroup$
The R-package GoodmanKruskal has a number of association coefficients for categorical data, which are well explained here (Cramer's $V$, Goodman and Kruskal's $\tau$).
Note that a binary variable is a special case of a categorical variable.
