Demand Management of Station-Based Car Sharing System Based on Deep Learning Forecasting

Algorithm 1

Input Record on number of pick-ups in training dataset

    Record on number of drop-offs in training dataset

    Record on time-of-day

    Record on day of week in training dataset

    Record on number of weather variables , , ,

    Look-back windows , , and

Output LSTM with learnt parameters

1:  Procedure LSTM training

2:    initialize a null set:

3:   for all defined time slice do

4:    

5:    

6:    , where

7:   , where , where , , , are the sets of different categories of explanatory variables in one observation

8:    A training observation is put into

9:    A training observation is put into

10: end for

11:   initialize all the weighted and intercept parameters

12:   repeat

13:    randomly extract a batch of samples from

14:    estimate the parameters by the minimization of the objective function shown in Equation (30) within

15:  until convergence criterion met

16: end procedure