Information value is one of the most useful technique to select important variables in a predictive model. It helps to rank variables on the basis of their importance. The IV is calculated using the following formula
IV = (Percentage of Good event - Percentage of Bad event) * WOE, where WOE is weight of evidence
WOE = log(Percentage of Good event - Percentage of Bad event)
Here is what the values of IV mean according to Siddiqi (2006)
If information value is < 0.03 then predictive power = "Not Predictive"
If information value is 0.03 to 0.1 then predictive power = "Somewhat Predictive"
If information value is 0.1 to 0.3 then predictive power = "Meidum Predictive"
If information value is >0.3 then predictive power = "Highly Predictive"