Logistic regression is a variant of ordinary linear regression used when the dependent observations can attain two values, usually the occurrence or non-occurrence of some event. The logistic model predicts the probability of occurrence as a function of the independent variables. Logistic regression fits the data with a special s-shaped curve using linear regression for the dependent variable and transforming it with the function , which produces p-values between 0 and 1. All of the restrictions and recommendations made for linear least squares regression apply to logistic regression as well.