Proability of failure

[josh_psic]

Hello,

I have a date set (n=150) with 2 variables: contract value ($) and a binary variables success/failures that indicate whether the contract was completed without problems.
I want to know, for a given contract value, the probability of failure. I decided to create a basic histogram with my data and use a cumulative distribution function.
What do you think of that? I would like to use a simple solution.

Thanks!

 

[GretaGarbo]

Try a logistic regression model with 0 or 1 (failures or success) as the dependent variable and the contract value ($) as the explanatory variable (" the x-variable).

 

[hlsmith]

As for @GretaGarbo recommendation, make sure you check the linearity of the logit assumption in logistic regression.

 

[Dason]

What software are you using?

 

[josh_psic]

Hello,

Thanks a lot for your help. The logistic regression works well! @Dason, I used excel with "RegressIt" free add-in.

 

[hlsmith]

Glad to hear you had 'success'! This will generate predicted probabilities, which can be used to interpolate values of your continuous variable within the range included the sample.