canonical correlation and bartlet's test for the discriminant analysis function

_ _Hello, i would like to carry out a discriminant analysis upon a set of variables [D] and a response [d].

cls = fitcdiscr(D,d, 'FillCoeffs','on')

How can I assess how well the model is, and the significance of each of the variables?. Some statistical programs provide both the canonical correlation of the model, and the p-value of each variable. Surely Matlab also does, but i have not already found out how. I've made the canonical correlation between data (D) and response (d), but i think this is not what i need since it is not taking into account the discriminant process:__

[A,B,r,U,V,stats] = canoncorr(D,d)

I would appreciate any help.