Algorithm 1 Trajectory Shapley

Input:   Randomly selected background sample $S=\{{\mathbf{G}}_1,\cdots ,{\mathbf{G}}_{\left|V\right|}\}$,   Explain region coordinate (x, y) in terms of grid,   Set trajectories $\mathbb{P}=\{{\mathbf{T}}_1,\cdots ,{\mathbf{T}}_{\left|N\right|}\}$ in t time slot,   Flow tensor $G_t$, pretrain model f.  Output:    1:Trajectory Shapley set M    2:Initialize $\varphi \left(G_t\right)=0$, $M=\varnothing$    3:Generate output ${\mathcal{Y}}_{t+1}=f\left(G_t\right)$    4:Obtain $\mathbb{E}\left[{\mathcal{Y}}_{t+1}\right]=\frac{1}{\left|V\right|}{\sum }_{G_i\in S}{G}_i$    5:Calculate $\varphi \left(G_t\right)$ using Equation (3)    6:for all trajectories ${\mathbf{T}}_{\mathbf{i}}$ in $\mathbb{P}$ do    7: Calculate $\varphi \left(T_i\right)$ using Equation (4)    8: Add $\varphi \left(T_i\right)$ to M    9:end for  10:return M