Filter by factors:
Scale by metric:
Let's breakdown the collisions by the factors that caused them. The data used includes collision data from the last 3 years. The leading factor related to collisions are dark conditions and wet roads. However, something to be careful is that it's a lot more common for roads to be wet or for it to be dark outside than for people to speed or drive under the influence. A much better measure would be out of people speeding or driving under the influence, how many of them get into an accident? Unfortunately, this is nearly impossible data to obtain because there's no way to track this information.
Additionally, there are some data quality issues because a lot of the collisions reported by SDOT don't document what the road conditions, speeding, alcohol, or light conditions were. This could mean that a lot of undocumented collisions were caused by one of these factors. The main takeaway here? These are all factors that drivers should be especially careful in, but darkness and wet roads have caused the most collisions out of the ones we have data for.
Risk by Factor
Using the same data as the donut charts, we now will compare factors based on whether they contained an injury, fatality, or neither. This was done through checking if an accident contained 1 or more injury, serious injury, or fatality. Note by doing this, we do not factor in accidents which contained multiple injuries or fatalities.
As shown, injury and fatal accidents are more likely to occur at night time, from alcohol, from wet roads, and by speeding. Unsurprisingly, they are less likely to occur due to parked cars. Again, there may be some data quality issues as mentioned in the previous section.
Here, we explore the number of collisions per month over the past 3 years. The overall counts and trends are about the same; it appears that the number of collisions slightly went up between 2014 and 2015, but slightly dropped in 2016.
The more interesting aspect to look at is the trends across months within a given year. We can clearly see that peak time of year for collisions occurs around September/October. The beginning of fall (and the poor weather associated with it), as well as things like the starting of the school year for many, are likely factors for this peak collision time.
Select portions of the axes using click and drag to filter by the various categories. Use the radio buttons to color by different categories. To remove a filter, drag it to the bottom of an axis, and drag the top of the bar to the bottom.
This visualization consists of a transparent dot for every collision in Seattle in the last 5 years. Each collision is a dot, and each dot's area represents the number of injuries that occurred in that collision. You can hover over a category to see the general distribution of that category; collisions not in that category will have a transparency applied.
Try looking at collisions involving right turns, and then look at collisions involving left turns. Do you see the difference?