Algorithm 6: RACC.

Input: Training data ${\left\{\left(x_i,y_i\right)\right\}}_{i=1}^n$, new data $x$, type of base classifiers $T$, subspace distribution $D$, integers $B_1$ and $B_2$, criterion $C$. Output: Predicted label $C_n^{RACC}(x)$, the chosen proposition of every feature $\eta$.  1. Generate random subspaces independently, $S_{jk}~D,1\le j\le B_1,1\le k\le B_2$.  2. For $j\leftarrow 1$ to $B_1$ do.    Choosing of optimal subspace $S_{j*}$ is performed from ${\left\{S_{jk}\right\}}_{k=1}^{B_2}$ based on $C$ and $T$.    End  3. Develop the collective decision function as an ensembled one, and represent it as: $$v_n(x)=\frac{1}{B_1}{\displaystyle {\sum }_{j=1}^{B_1}{C}_n^{S_{j*}-T}}\left(x\right)$$ .  4. Based on Equation (2), the threshold is set.  5. The predicted label $C_n^{RACC}(x)=1\left(v_n(x)>\widehat{\alpha }\right)$ is given as output, which is the chosen proposition of every feature $\eta ={\left(n_1,\mathrm{\dots },n_q\right)}^T$.