UM-3 : Multiple Linear Regression

Abstract

The derivation of relationships among hydrologic variables is of great importance for the transfer of information from few gauged sites to many ungauged sites. This relationship may be between one dependent variable and one or more independent variables. The general form of the multiple linear regression is:

X 1 where:

,B + B2 + e 1 X2 + B3 X3 +Bm Xm

X1

is dependent variable

are independent variables and

X2 , B1,

,Xm

,Bm

are regression coefficients

The user's manual gives the details of the computer programme for multiple linear regression. The selection of the different sets of independent variables and designation of a dependent variable can be made as many times as desired.

The output of the programme includes means and standard deviations of dependent and independent variables, correlation coefficients between independent variables and dependent variable, regression coefficients, standard error of regression coefficients, computed t-values, intercept, multiple correlation coefficient, standard error of estimate, analysis of variance for multiple regression and table of residuals.

The manual describes various statistics given in output with test example and input data specifications. The programme is written in FORTRAN . The manual also gives the limitations of the programme.