Forecasting passenger demand is of great interest for public transport operators. Despite the important role that forecasting play in mobility demand understanding, in-depth transport oriented analysis of the forecasting results is often overlooked, since it raised some challenges. In this context we developed two visualization tools with open source frameworks that allow to analyze spatio-temporal time-series forecasting with context awareness. The first visualization tool allows to analyze the forecasting results over large period in all the stations and to zoom in for more precise temporal details. The other tool allows to better understand the passenger demand relations between the different stations of the transport network, and enable a spatial analysis of the results. Analyzed time-series corresponds to the forecast results of the number of passengers entering each station with a fine-grained temporal resolution (15 minutes interval) during one year achieved with a well-known machine learning model, a Random Forest. In order to highlight the spatio-temporal specificity of the passenger demand, we have computed and analyzed the residuals of a long-term forecast model that returns normal passenger demand. Here we show that both visualization tools depict the stations and the period hard to predict and allow to have an insight on which contextual element (weather, event on the city and incident on the transport network) could impact the forecasting. Experiment are performed with real data given by the transport authority of Montreal (Société de transport de Montreal, STM).