Spatiotemporal data mining: a survey on challenges and open problems

Abstract

Spatiotemporal data mining (STDM) discovers useful patterns from the dynamic interplay between space and time. Several available surveys capture STDM advances and report a wealth of important progress in this field. However, STDM challenges and problems are not thoroughly discussed and presented in articles of their own. We attempt to fill this gap by providing a comprehensive literature survey on state-of-the-art advances in STDM. We describe the challenging issues and their causes and open gaps of multiple STDM directions and aspects. Specifically, we investigate the challenging issues in regards to spatiotemporal relationships, interdisciplinarity, discretisation, and data characteristics. Moreover, we discuss the limitations in the literature and open research problems related to spatiotemporal data representations, modelling and visualisation, and comprehensiveness of approaches. We explain issues related to STDM tasks of classification, clustering, hotspot detection, association and pattern mining, outlier detection, visualisation, visual analytics, and computer vision tasks. We also highlight STDM issues related to multiple applications including crime and public safety, traffic and transportation, earth and environment monitoring, epidemiology, social media, and Internet of Things.

Keywords: Challenges Issues; Data Mining; Research Problems; Spatial; Spatiotemporal.

Fig. 1

Spatiotemporal data types. a spatiotemporal events of different types at different locations and timestamps. b spatiotemporal trajectories between locations ($l_1$ and $l_n$) at time ($t_1$ and $t_2$). (c and d) spatiotemporal point reference data at different locations at timestamps ($t_1$ and $t_2$). (e and f) spatiotemporal raster data of regular grid at time ($t_1$ and $t_2$)

Fig. 2

A taxonomy of the proposed STDM challenges structure. The survey is designed to cover the STDM related challenges from three different perspectives. We propose to investigate the general challenges that affect the STDM in terms of relationships, data, natures and limitations of research. Then, we discuss the STDM tasks and applications focusing on their related challenges

Fig. 3

A word-cloud visualisation of the most frequent used search keywords

Fig. 4

Related work distributions for journal articles and conference proceedings

Fig. 5

Related work distributions for years from 2011 to 2020

Fig. 6

Cause-and-effect diagram of STDM general challenges. The figure shows a taxonomy of STDM challenging issues

Fig. 7

Examples of topological relationships between two areas

Fig. 8

Different sources of data needed for crime analysis

Fig. 9

Spatial scaling between a or b and c and zoning between a and b. The figure shows the impact of having different scales and zones on the analysis results

Fig. 10

Vagueness due to data similarities stem from different criteria. Trajectory 2 and 3 have similar spatial attributes. However, they are semantically different

Fig. 11

Dynamic changing of the spatiotemporal distribution of moving objects

Fig. 12

An example of explaining a model prediction of flue based on different symptoms, from LIME Ribeiro et al. (2016)

Fig. 13

Three different types of outliers Ji et al. (2019)

Fig. 14

Drone-based object tracking (Hamdi et al. 2020b)

Fig. 15

An example of semantic amodal visual object segmentation Zhu et al. (2017). The first row shows the original scene and its segments human-annotation. The second row visualises the depth and visible edges. Finally, the third one shows the semantic annotation of the invisible regions

Fig. 16

The impact of different correlations among regions on RLRH demand forecast. For example, R7 is adjacent to R8, similar to R4 and R2, connected with R3, and distant or irrelevant to R6

Fig. 17

Survial analysis under uncertainty (Sokota et al. 2019). The survival curve, red line, calculates the probability as a temporal function. The point-wise and simultaneous intervals covers the uncertainties

Fig. 18

Mobile-collected data can be used to monitor different patterns of user’s walking, voice, tapping, and memory (Schwab and Karlen 2019)