Article Preview
TopIntroduction
With the development of Internet technology and the popularization of location-based services, researchers have discovered valuable knowledge through the analysis and mining of urban S-T big data, which has promoted people's lives, improved urban operation efficiency, reduced resource consumption, and achieved sustainable urban development (Wang et al., 2020; Li. 2021; Cheng et al., 2019). In recent years, significant advances have been made in artificial intelligence-based algorithms, which can make accurate predictions of complex problems with minimal domain knowledge and strong generalization ability, and AI-based algorithms have found a wide range of applications. These characteristics lay a foundation for studying traffic flow (TF) prediction (Li et al., 2021; Shanshan et.al., 2022). As an important part of urban S-T big data mining, TFP plays a key role in developing cities and intelligent transportation. Transportation, a field crucial to daily life, can help promote high-quality social and economic development. To achieve safe, comfortable, convenient, and green transportation is a key link to improving people's sense of security and happiness and ensuring social stability (Liu et al., 2020; Huang 2019; Nguyen et al., 2020).
The surge in private cars will inevitably lead to traffic congestion. The toxic substances emitted by vehicles in the process of congestion will not only damage people's health, induce various diseases, and affect life and work efficiency but also cause issues such as environmental pollution, resource waste, and economic losses (Zhang et al., 2019; He, 2020; Zhang et al., 2020). Accurate TFP can help analyze road planning, recommend more intelligent travel routes, and reduce traffic accidents. At the same time, if we can realize the urban TFP we can provide a reference for residents to travel, avoid hot traffic areas, ease traffic pressure, reduce traffic control pressure, and improve travel efficiency (Cheng et al., 2021; Cui 2020; Wang et al., 2020).
The parameter method with statistics is a general TFP method, and the Auto-Regressive Integrated Moving Average (ARIMA) model is a typical parameter method. In the 1970s, Ahmed and Cook (1979) first used ARIMA in TFP field to predict short-term TF of the expressway. Later, scholars put forward a variety of improved models, such as KARIMA (Voort et al., 1996), SARIMA (Williams & Hoel, 2019), and STARIMA (Kamarianakis & Prastacos, 2003).