Mobility services require accurate demand prediction both in space and time to effectively accomplish fleet rebalancing, present efficient on-demand transportation services, and allow for advanced ride-sharing with minimum fleet size. Although the optimization of mobility services is a widely studied topic, the critical demand prediction component has received less attention. In this paper, we aim to develop an efficient method for traffic demand forecasting by means of deep learning and hierarchical reconciliation approaches. The concepts, as well as the theories behind the proposed approach, are founded on a Hierarchical Time Series (HTS), which also adopts Long Short-term Memory (LSTM) as a special kind of Recurrent Neural Network (RNN) for the deep learning of the associated time series and producing reliable demand predictions. Herein, the proposed approach relies on the proper design of the HTS structure to find coherent forecasts for the number of trip departures and its associated uncertainty over predefined zones as well as over aggregated collections of these zones for applications like mobility service operations. Moreover, an error analysis is essential for accomplishing the reconciliation in the HTS structure optimally. The three main stages of the proposed approach (i.e., deep learning, error analysis, and optimal reconciliation), which independently function within the approach structure, have a remarkable ability to predict the demand, control all the forecasts at all levels of the hierarchical structure, and finally lead them to their coherent estimates. We evaluate the proposed approach on a large-scale GPS tracking dataset of Lyon in France. The proposed method reduces the root mean square error (RMSE) by 13.92% and 14.77% for the predefined and aggregated zones, respectively, compared with the LSTM using the historical demand and the external features of time at fifteen minutes time resolution. Similarly, the corresponding improvement for mean absolute percentage error (MAPE) is 14.87% and 19.23%, respectively.