UP - logo
ALL libraries (COBIB.SI union bibliographic/catalogue database)
  • Generalized multiple linear regression (GMLR) using factor analysis
    Kavkler, Ivan
    The paper discusses an example of generalized multiple linear regression (GMLR) by using factor analysis. In many cases explanatory variables are correlated, which prevents the use of classical ... multiple linear regression (MLR). By using factor analysis this problem is avoided because explanatory variables are replaced by a smaller number (m) of uncorrelated common factors, which allow the correlations between explanatory variables to be considered. Factor analysis allows simple calculation of generalized regression coefficients. An applications of GMLR is demonstrated on a case of "Minesweeper" computer matrix game. We have been able to create a random sample (N = 150) by actually playing the game, and have achieved, under similar conditions, a satisfactory approximation of normal distribution. In order for this method to ce complete, it requires a further study of explained variance and, because the explanatory variables are correlated, alsoa study of explained covariance. This is a matter of further study.
    Type of material - conference contribution
    Publish date - 2001
    Language - english
    COBISS.SI-ID - 5562396