Regression, in statistics regression means modeling and analysis of variables. It shows the relationship between a dependent variable and a independent variable, it may be one or more then one. Now we more simplify the term regression, it shows how a dependent variable changes when we vary one independent variable and treat another independent variable as a constant. Regression is used for prediction and forecasting. It also shows how an independent variable is related to dependent variable.
What is stepwise regression?
Stepwise regression means regression analysis in which the choice of predictive variable is taken out by an automatic procedure. If we more specify step-wise regression then suppose we have any mathematical equation with several dependent and independent variables which we have to solve by step-wise regression modeling. By applying several techniques we first eliminate the independent variables one by one to find the final value.
For better understand the step-wise regression we have to take the stepwise regression example, because with example we easily understand the procedures. For this we take a quadratic equation such as (aX^2+bX+c). In this we have to find dependent variable X by eliminating the independent variables. In step one; we separate the middle term in two parts with checking suitable multiplication of first and third term and addition also. In step two, we take out the common terms. In step three, we equate each term with zero and then finally we get the value of X.
Step-wise regression is of various types. First we discus about forward step-wise regression, this type of regression starts with no variables in the model. Basically we have to improve the model in forward stepwise regression by applying suitable procedure, such ass apply addition of each variable from a given model and improve the model. Repeat the procedure until none improvement take place. The total calculation is based on the F-distribution and degrees in the regression terms.
Now we discuss about another type of stepwise regression which is stepwise linear regression. This type of regression shows the relationship between a dependent variable which is scalar and one or more explanatory variables. If one explanatory variable is used then simple regression and more than one explanatory variable is known as multivariate linear regression, where the multiple variables are predicted which correlated with dependent variable. In linear regression linear predictor functions are used in regression analysis. Linear regression has also many practical uses.
