The author reminds the definition of coefficient determination and the idea of the largest credibility method in the introduction of the article. Firstly, the aspect of new coefficient defined by McFadden as a power of prediction process and called pseudo was characterized. The similar type of coefficients proposed by other known statisticians were described. Moreover, the hierarchic way of building the logistic regress model, through adding next variables to adjust an estimated model to empirical data was introduced. Four kinds of variables were analysed, whose adjustment influence on the estimated logistic regress model was affected by the quantity of inconsistences which appeared in the data result. The calculations of parameter made by the McFadden’s, Nagelkerke’s and Cox-Snell’s formulas were presented. Concise conclusions relating to the estimated logistic regress model on the basis of empirical data from tested MD-8 fuses type were introduced in the end of the article and this model was compared to values of coefficient counted in the article. It was stated, that McFadden’s pseudo parameter is most often used and it defines the power of prediction process.
- coefficient of determination (R-squared)
- logistic regress
- power of prediction process