Technological advancements are transforming city operations and urban planning. The need for timely and accurate tourism data has never been more critical, as highlighted by the challenges faced during the COVID-19 pandemic. Despite advances in big data, applications of tourism demand forecasting in the literature have been slow to adapt to a data-rich world. Traditional methods in tourism demand forecasting often rely on lagged survey data analysed at quarterly or annual intervals. This research introduces a novel framework for hourly, segment-based tourism demand forecasting at the suburb level across a destination. This approach leverages of spatiotemporal mobile phone data collected through a telecommunications network over a four year period. This work is made possible through a collaborative agreement with the City of Gold Coast, and highlights the potential for smart cities to harness big data for urban planning and tourism management. Our study employs both local statistical time-series models and global neural network techniques to evaluate forecasting performance, aiming to bridge the gap between complex modelling techniques with their practical application. We discuss some practical forecasting challenges and data science considerations pertinent to high-frequency tourism data usage including addressing complex seasonality, multi-step forecasting, and performance monitoring and model updating to combat concept drift. This study focuses on the role of data science in transitioning from traditional forecasting approaches to meet the demands of the big data era within smart cities. This research appeals to a broad audience, from forecasters and tourism academics to industry practitioners, all of whom benefit from deeper insights and actionable intelligence in a rapidly evolving landscape.