Algorithm 8: The procedure for applying the DT diabetes detection model

Input: Input feature vector with n-samples and d-dimension $X\in {\mathbb{R}}^{n\times d}$ and true      label $Y\in {\mathbb{R}}^{n\times 1}$     Result: The posterior $P\in [0,1]$       1 Divide $\theta =(j,t_m)$ into $Q_{left}\left(\theta \right)$ and $Q_{right}\left(\theta \right)$ subsets, where $\theta$ consisting of a feature, j and threshold, $t_m$.       2 Use an impurity function (H), which are given below, to calculate the impurity at the $k^{th}$     node, $G(Q,\theta )=\frac{n_{left}}{N_m}H\left(Q_{left}\left(\theta \right)\right)+\frac{n_{right}}{N_m}H\left(Q_{right}\left(\theta \right)\right)$, where     $H={\sum }_c{P}_{mC}\times (1-P_{mC})$ or $H=-{\sum }_c{P}_{mC}\times log\left(P_{mC}\right)$ and     $P_{mC}=\frac{1}{N_m}{\sum }_{x_i\in {\mathbb{R}}_m}I(y_i=C)$       3 Reduce the impurity by selecting the parameters, ${\theta }^{*}=argmi{n}_{\theta }G(Q,\theta )$.       4 Reapply the preceding steps for subsets $Q_{left}\left({\theta }^{*}\right)$ and $Q_{right}\left({\theta }^{*}\right)$ until depth reach to     $N_m<samples\left(minimum\right)$ or $N_m=1$.