EViews.com

Hey there!

I'm currently trying to discriminate a set of variables, comprising 5 categories.
Within each category, I have several ratio's, subdivided in two categories (0 or 1 as a binary indication) But at the end of the day, I only need 1 ratio within each category, that discriminates the best between 0 and 1.

My question is the following : What is the best and most accurate way to see how discriminating variables are ?

Thank you in advance!

I don't understand the question.

Let's take an example :

I have a group of entreprises, which are either in default or not.
To define those entreprises, I have for example 2 ratio's.
What I want is, by using Eviews, to know which ratio discriminate the best among defaulted and non defaulted firms.
In other words, how can I see which ratio differentiate the best firms that are in default and those that are not.

Here is a schema :

Ratio 1 : Defaulted : 4 5 4 4 2 4 5 7 6 4 5 5 4 2
NON defaulted : 4 1 2 5 4 3 2 3 4 5 8 6 2 4

Ratio 2 : Defaulted : 7 8 5 9 7 4 8 9 6 8 9 9 7 8
NON defaulted : 1 2 0 0 1 2 -1 0 2 1 0 0 1

Here, I can see that ratio 2 differentiate the best data from defaulted firms and data from non defaulted firms.

Run a probit or logit using 1 for default and 0 for not as the left hand side variable and the two ratios as right hand side variables.

That is my final goal, but actually I have several candidates for each variables. In the case where :

Y(1 or 0) = b0 + b1X1 + b2X2 + b3X3, I have 3 candidates for X1, 3 candidates for X2 and 4 candidates for X3. I can run several logit regressions and see in the end which one is the most accurate but that would be a lack of time since I know it is possible to "test" variables side-by-side and see which one is the most discriminative between defaulted and non defaulted firms..Anyone?