Space-time Crash Analysis

 Leeds  is an urban area located in the metropolitan county of West Yorkshire, United Kingdom. The city is situated northeast of Manchester about 50 kilometers and is composed of 33 wards that lie along the River Aire. The  total population  of Leeds as of 2019 rests around 818,085 inhabitants and has exhibited a steadily increasing growth rate through the years, most recently calculated at 1.08% in 2019. The city of Leeds is recognized as the  fifth-largest city  in the United Kingdom with a population density of around 1,416 people per square kilometer. A comparison map was created below illustrating the change in population density between years 2009 and 2018. Move the interactive swipe tool to the left and right to peer the changes over the ten-year period.

 Road casualties  cause an estimated 1.2 million deaths and 50 million injuries every year around the world. According to the  Leeds City Council , the urban area averaged around 2000 to 3000 vehicle accidents over recent years. An animation was created to illustrate crash point locations in Leeds, U.K. between 2009 and 2018. A trend can be noticed that most of the points tend to aggregate around the city center of Leeds.

The analysis began with space-time pattern mining, which involved compiling crash points into space-time hexagon bins where the points were then counted and aggregated by time-step and distance intervals.

• Distance intervals determine the size of the bins and are used to aggregate the crash points.
• A single time-step represents the amount of time, and thereafter, the distance and time intervals are aggregated into the bins.

To explore the traffic accident trends between 2009 and 2018, we looked at 200-m hexagons using 16-week time periods, which produced a 0.83% bias in a NetCDF file. To display the space-time hexagon grid in a meaningful way, emerging hot spot analysis was performed which involves identifying crash point density trends and is illustrated by the map below.

Leeds transportation networks have a prominent reputation in the United Kingdom, as the railway station is one of the busiest in Britain. Despite that,  Hyde (2019)  proclaims that the transport network in Leeds at its current state is unsatisfactory, complaining of its excessive traffic jams, delays and unreliable public transport services. The Leeds City Council has  joined forces  with the West Yorkshire Authority to invest in improving the transportation infrastructure and are looking to target the most troublesome areas first.

The number of accidents in each ward were separately summed and normalized by the area of the administrative unit, which is called 'accident density'. The accident density maps of Leeds, United Kingdom between 2009 and 2018 were assembled into an animation using 5 classes determined by the natural break of the first year.

Kernel Density Estimation (KDE) is a non-parametric way of estimating the probability density function of a random variable. This estimation technique deals with the fundamental data smoothing problem, where inferences about the population are made, based on a finite data sample. The following map showing the kernel density of accident hot spots in Leeds from 2009 to 2018.

After exploring the data, the next step in the analysis was to associate each crash with the underlying road network and look for crash hot spot locations. First, the crash points between 2009 and 2018 were snapped to the road segments deemed closest. The crash point table and road network table were then spatially joined by number of accidents as a summed per road segment. Then the crash rate was calculated into a new field representing the average cars wrecked per kilometre per year, and was calculated as follows:

• n = number of cars wrecked per road segment
• y = number of years (10)
• l = length of road segments

Though the overall number of road casualties has fallen over the last few consecutive years,  Leeds City Council  reports the number of people killed or seriously injured has risen 4%, hence the figure is still unavoidable. Therefore, identifying areas with the highest concentration of accidents is advantageous for minimizing road accidents related issues and taking proper initiatives for future traffic safety planning, not only for Leeds but for other geographic regions as well.

The purpose of the comparison is to note the differences between the data mining and data analysis outputs.  Data mining  is meant for exploring the data and any trends it may portray before analysis.  Data analysis  involves modelling the data to create meaningful visualizations, bringing useful information that can aid in decision-making.

The Leeds City Council has joined forces with the West Yorkshire Authority to invest in improving the transportation infrastructure and are looking to target the most troublesome areas first. Per  Hyde’s 2019  article, studies from the council suggest 52 percent of Leeds-bound trips are made by car, which may support why high volumes of car crashes occur each year in the city.

In general, among the people involved in road accidents, the number of men is higher than females, and 80% of them did not experience any serious injury. However, most of the accidents took place in and around the city center, such as Holbeck, New Wortley, Headingley, and Patternuton, that are worth mentioning. Moving from the city center to the periphery, one can see both the number and density of accidents start to decrease. The city hub and its surrounding areas had experienced around 85 accidents, whereas the outskirts around 3. Based on this analysis, it can be concluded that the downtown areas are the hot spots of road causality in Leeds, as most of the accidents happened within these areas. The city center should be targeted by the Leeds City Council for infrastructure improvement.