Table 2 Comparison of continuous-variable/discrete discriminant analyses
Predictor, a | Classes | Parameters | Predictor distribution for class y | Prediction |
|---|---|---|---|---|
Continuous variables | Case (y=1) | μ 1, Σ 1 | N(μ y ,Σ y ) | Decision boundary: |
|  | Control (y=0) | μ 0, Σ 0 |  | δ 0(a)=δ 1(a) |
Discrete variables | Case (y=1) | \(\psi _{1}=\{h_{i}^{(1)},J_{ij}^{(1)}\}\) | Pr(a;ψ y |y) | \(\beta _{i}=h_{i}^{(1)}-h_{i}^{(0)}\) |
| Â | Control (y=0) | \(\psi _{0}=\{h_{i}^{(0)},J_{ij}^{(0)}\}\) | given by Eq. (3) | \(\gamma _{ij}=J_{ij}^{(1)}-J_{ij}^{(0)}\) |