Transform a categorical variable

this stand-alone transforms a categorical variable into a quantitative variable using Fishers method

The response variable is discrete ( it could be binary or multinomial), it could have "k" groups (k=2 binary, k>2 multinomial). the categorical can have "c" categories. The routine finds the weights for each category using the Fisher's method described in
Hubert J. Carl, Wisenbaker M. Joseph, Smith D. Jerry and Smith C. Janet (1986). "Using Categorical Variables in discriminate analysis', Multivariate Behavioral Research, 21:4, 479-496

This routine is useful if you are performing a regression or ANOCOVA or using a GLM but you have many categorical variables and there are many categories that dummy coding them is inefficient. however, the practicality of the algorithm is not known.