Selecting an appropriate model to forecast product demand is critical to the manufacturing industry. However, due to the data complexity, market uncertainty and users' demanding requirements for the model, it is challenging for demand analysts to select a proper model. Although existing model selection methods can reduce the manual burden to some extent, they often fail to present model performance details on individual products and reveal the potential risk of the selected model. This paper presents DFSeer, an interactive visualization system to conduct reliable model selection for demand forecasting based on the products with similar historical demand. It supports model comparison and selection with different levels of details. Besides, it shows the difference in model performance on similar products to reveal the risk of model selection and increase users' confidence in choosing a forecasting model. Two case studies and interviews with domain experts demonstrate the effectiveness and usability of DFSeer.

Fig. 1. Analyzing the influence of road topology and traffic conditions on POI accessibility: density distribution before (a) and after (b) removing the selected road segments FE and ED; and density distribution before (c) and after (d) slowing down the average speeds of the selected road segments MN and NO.

Nowadays, the population has been overgrowing due to urbanization, yielding many severe problems in the urban area, including traffic congestion, unbalanced distribution of urban hotspots, air pollution and so on. Due to the uncertainty of the urban environment, it always needs to integrate experts' domain knowledge into solving these issues. In recent years, the visual analytics method has been widely used to assist domain experts in solving urban problems with its intuitiveness, interactivity and interpretability. In this survey, we first introduce the background of urban computing, present the motivation of visual analytics in the urban area and point out the characteristics of visual analytics methods. Second, we introduce the most frequently used urban data, analyse the main properties and provide an overview on how to use these data. Thereafter, we propose our taxonomy for visual analytics in the urban area and illustrate the taxonomy. The taxonomy provides four levels for visual analytics on urban data from a new perspective based on the four stages in data mining.Four levels from our taxonomy include: descriptive analytics, diagnostic analytics, predictive analytics and prescriptive analytics. Finally, we conclude this survey by discussing the limitations of the existing related works and the challenges to visual analytics in the urban area.data mining, urban data, visual analytics, visualization | INTRODUCTIONRecently, the demand for building smart cities has been accelerated by the desire for high-quality daily lives for human beings and technological development. However, urbanization causes many human behaviours in the urban area and generates multi-source heterogeneous urban data.Meanwhile, an increase in the population can also lead to many severe problems, including traffic congestion, unbalanced distribution of urban hotspots, air pollution and many other domains. Paulos and Goodman (2004) first proposed the term 'urban computing', and many researchers have moved into this area so far. For example, in previous research, people can use the GPS trajectory data collected from vehicles to analyse traffic conditions (Xin et al., 2011) and keep safe traffic control (Abbasi et al., 2020). Furthermore, urban planners adopt cell phone data (Cho et al., 2011) collected from the base stations to explore the human mobility problem and optimize the location of urban hotspots. They also use air quality data collected from many sensors located in an urban area to analyse the air quality (Deng et al., 2019;Harbola et al., 2021). Moreover, such data can also be used on public health analysis (Antweiler et al., 2021), transit route planning, crime analysis (Svicarovic et al., 2021) and decision-making for emergency response (Johnson & Jankun-Kelly, 2021). This is beneficial to the development of a modern city and raises the happiness quotient.