Traffic Accidents in Helsinki
This visual data mining case study is based on open data available at the Helsinki Region Infoshare. The original data set contains information about traffic accidents in Helsinki during 2000-2010. Several visualizations based on the same data set has been published earlier by Finnish media (see MTV3.fi, HS.fi and SuomenKuvalehti.fi) but surely there is still room for one more try...
In this study, we took the coordinates of all accidents and visualized the density of these points as a heatmap on Google Maps. The heatmap was generated using the HeatMiner® visual data mining tool. Heatmap colors indicate high density regions so that red areas have the highest accident density.
In fact, 10% of all traffic accidents in Helsinki happen in the red area. The orange and red area together cover 20% of the accidents, the yellow area adds 10% more and so on. The last blue step adds 25%, so in total the colored areas covers 75% of the accidents. Accidents happen also in non-colored areas but their density is too low to be particularly interesting.
As you can see, heatmaps are a great tool to summarize large amount of data points (29142 in this case). If you were driving a car in Helsinki, a single glance to the heatmap would tell you which areas to avoid. Similar instant insight is hard to achieve with other type of visualizations such as scatter plots (points on the map).
For example, take a look at the picture below showing some traffic accident locations. Are there more blue points than red points or vice versa?
Most of us would say that blue points are clearly dominating. But actually there is an equal number (3000) of red and blue points! The red points are just so close to each other that they overlap in the scatter plot which gives a totally false visual impression.
- Video interview of Pauli Misikangas in which he gives insights into this case study and describes the benefits of the chosen approach (in Finnish)
HeatMiner® is a trademark of Agience Oy Ltd