A look at Street Safety in Cambridge, MA

Why we should care — What it looks like today — Where it’s heading — What we can do about it

Street safety is an obvious concern. No matter what we do, where we live, or how we get around, everyone cares about the safety of the streets where they live. But it wasn’t until my recent participation in Participatory Budgeting that I realized how crucial street safety is in local government.

So I decided to dive into the data around street safety. I wanted to find out more about things like where did crashes take place? Or how do crashes involving bicycles and pedestrians compare? And whether or not our street are getting safer? What I’ve written below are the results.[1]

Why we should care

The number one category of the number of ideas submitted in Participatory Budgeting in Cambridge, Massachusetts was Streets, Sidewalks & Transit. It is overwhelmingly more popular than other local concerns. This could be due to how residents view Participatory Budgeting and what things they might think are best to tackle within that budget, but it is revealing in that it expresses an area of concern for residents that is pretty approachable and solvable without requiring massive capital investment. Something all government officials should enjoy.

Another reason why we should care about street safety is because of the goals Cambridge has set out for itself. Particularly, its Vision Zero goal, the goal to eliminate all traffic fatalities and severe injuries. The city has laid out a plan for this vision but there is still a long way to go before achieving it.

Finally, we should care about street safety because city populations continue to grow and show no signs of slowing down. In addition, there are more and more types of vehicles appearing on our streets, including electric bikes, scooters, skateboards, and even autonomous cars. If we don’t figure out better ways to improve our roads now, it’s going to only get increasingly challenging to do so in the future.

What it looks like today

Now that we understand how important it is to improve our streets, we need to get a better sense of what the current level of street safety is today. In my opinion, one of the best ways to understand city street data is through visualization. Nothing drives a point home the way looking at the data on a map does. That’s why I put together this map of all the crash data in the city of Cambridge from 2010 to 2017.

Taking a look at the map, we can get a better sense of where crashes are taking place in the city.

All crashes in Cambridge from 2010–2017

You’ll notice a few things that clearly stick out on the map when looking at it. The first is that the major road arteries of the city (Mass Ave, Broadway, and Cambridge St) have the majority of crashes. It is pretty obvious why, because they also have the most traffic, but the thing that stands out is not what’s included but what’s not included — Memorial Drive.

Except for a few spots near major intersections or shopping areas, there are very few crashes that take place on Memorial Drive. Especially as compared to Mass Ave, which is almost entirely red (higher density of crashes) through the whole map.

This raises a few different questions. The first is that maybe crash reporting for Memorial Drive is inaccurate due to crashes being reported on nearby streets rather than on Memorial Drive? The could be due to drivers who are involved in crashes on Memorial Drive wanting to get off of it, in the interest of their own safety, before reporting it. Though if this were true, then we would likely see a high amount of crashes right next to Memorial Drive, which we don’t see in the data.

Another, more interesting question is whether the road composition has anything to do with the frequency of crashes? For example, Memorial Drive is a four-lane road (two lanes each way) that has few traffic lights, crosswalks, and intersections. Compare that to Mass Ave, which is a two-lane road (one lane each way, for the most part) and has a ton of traffic lights, crosswalks, and intersections.

Can the much lower frequency of crashes on Memorial Drive be explained by this difference? I’m not entirely sure. I would guess the amount of traffic traveled on each road every day is probably roughly similar, it’s even possible there is more traffic on Memorial Drive than on Mass Ave because of how slow and crowded it can get.

So it would seem that perhaps the amount of stopping and going, combined with having to watch out for bikes and pedestrians, makes it a much more crash-prone road to drive. This is an important point to remember as we look at the locations of crashes involving bikes and pedestrians and think about potential solutions to reducing their frequency.

Looking at bike related crashes, we can see that they almost exclusively are concentrated in 4 areas: Mass Ave (both north and south of Harvard Sq, and including Harvard Sq), Broadway, Hampshire St, and Cambridge St.

Bike related crashes from 2010–2017

This makes sense since they are the major roads for bikers to travel on. But again, it also shows the infrequency of crashes on major roads that have separated bike lanes from the road, like Memorial Drive and the Fresh Pond area.

The most interesting part of this data is the Inman Square area. This area has a high concentration of crashes involving bikers and appears to make up a good amount of all crashes in this area. This is definitely an area that has come under a lot of scrutiny from residents in the past but has yet to be addressed.

If we look at pedestrian related crashes, it is clear that the heavy pedestrian areas (Central Sq, Harvard Sq, and Porter Sq) have the most occurrences of crashes. What stood out to me in the pedestrian data was the concentration of crashes near the Fresh Pond Mall. I had previously been unaware that that area was hazardous to pedestrians but by looking at the data it appears to be one of the more crash-prone areas in Cambridge for pedestrians.

Pedestrian related crashes in Cambridge from 2010–2017

It’s also interesting to note the differences between locations of bike and pedestrian related crashes. Bike related crashes are much more spread out along streets, whereas pedestrian ones are very concentrated near major intersections. This might suggest the frequency of bike crashes with parked cars, from opening of doors for example.

Another thing I was curious about when trying to get a better understanding of the current state of crashes in Cambridge was when they were happening, so I looked at both the time of day and day of week. Predictably, it shows that rush hour traffic has the highest concentration of crashes, with the majority of crashes happening on weekdays and during the morning or evening commutes.

Crashes by Day of Week
Crashes by Time of Day

Where it’s heading

Naturally, after looking at data like this the question of whether or not things are getting better arises. In order to answer that question, we need to look at the frequency of crashes over time.

Below is a look at the number of crashes by month from 2010 to 2017. The highlighted section where there is a major spike in crashes is for February of 2015. This was when the Cambridge area was hit by a number of large snow storms, which incidentally even broke the snowfall total records. Despite the likely lower amount of traffic in the streets during these months, the people who did brave the weather ended up getting into a high amount of crashes.

All crashes over time by month

Besides the overconfidence of Boston area drivers in the snow, the graph above seems to show a general steady trend of crashes over time. To get a better sense of the trend of crashes over time, I plotted a linear regression model against the total amount of crashes over the years.

Trend of total crashes

As you can see from the graph, it pretty much shows an even trend of crashes, with it actually slightly trending up. This isn’t necessarily a bad thing since it doesn’t account for things like severity of crashes and crashes as a percentage of total traffic (maybe the total amounts of people and cars on the road are increasing even faster?), it does at least indicate that it’s an area with room for improvement.

And just to give you more of an idea about how bicycle and pedestrian related crashes are trending, I’ve included those graphs below as well.

Trend of bicycle related crashes
Trend of pedestrian related crashes

Again, don’t think that these graphs show a terrible trend for pedestrians and things getting better for bikers (but they might!), instead treat them as indications of where things may be going but also consider that this is just the trend of frequency of crashes, not necessarily the severity or proportion.

Another way we might look at the data is seeing how the frequency of crashes by objects involved look over the years. This chart is produced below.

Total crashes by year broken out by type

Again, it shows a pretty even trend over time for all crash types. I would guess that most of the discrepancy in the numbers over the years is in reporting differences rather than any policy changes but further exploration into this data would be needed in order to determine the effects.

What we can do about it

Whenever I write a post like this that indicates there may be things we need to improve or fix, I don’t like ending it without thinking about potential solutions. The good news for this particular issue is that there are actually a lot of pretty straightforward ways to improve road safety. In fact, the city of Cambridge has already begun some of them, it would just be better if they went about it a bit quicker. But I guess that’s government for you.

Anyway, I think there are four main ways to approach the issue of street safety in Cambridge. The first is to fix the known problems. If you’ve lived in Cambridge for a bit and haven’t been exposed to the atrocity that is the Inman Square intersection then lucky you! However, for the rest of us, we know how terrible it is. But so does everyone who has ever been through it: government officials, residents, and visitors. It’s been talked about for years but still is yet to be fixed or to break any ground on a new solution. Now’s the time to do it, if this data is any indication, fixing the Inman Square intersection could have a noticeable impact on crash frequency in Cambridge.

The next approach is to build more infrastructure. Again, the city has already begun on this work by starting to build out the network of protected bike lanes as laid out in its Bicycle Network Vision. Many of the proposed protected bike lanes are in the exact areas that are needed most, Mass Ave and Cambridge St. If the infrequency of bicycle related crashes on Memorial Drive is any indication, it is that separating auto traffic from bicycle traffic can have a large impact in reducing the number of crashes.[2]

And while we’re talking and analyzing data, the third approach is to use the data! While the data I’ve explored in this post is not perfect by any means, it definitely helps show generally where the problem areas are. There’s really no excuse not to use the data. If you know an intersection has a high concentration of pedestrian crashes than change the intersection! Add more signs, slow traffic down, install or remove traffic lights, add raised crosswalks, etc. There are a lot of ways to try and fix the issue, but ignoring and thinking it will get better it is not one of them.

Finally, let’s explore some new and different ideas. Beyond just fixing the obvious problems with the obvious solutions, I think we should start thinking about new solutions.

I recently read Charles Montgomery’s Happy City which outlines a few interesting ideas that I think Cambridge could experiment with to increase street safety in the city.

There are a number of different ideas laid out in the book, like days without automobiles, building more and better parks, and changing zoning laws, but the one I found most interesting and straightforward to implement was the idea of converting some sections of roadway into pedestrian areas. A great example of this, as pointed out in the book, is New York City’s experiment of closing off parts of Broadway to traffic.

As you might expect, there was a lot of initial push back from the public. But in the end, its been a resounding success. Not only did it actually reduce traffic in the area but it also had a sort of revitalizing effect on the immediate area by increasing pedestrian activity and creating new vibrant areas that benefits both businesses and residents alike.

So with that in mind, I’ve actually started my own petition to get the City of Cambridge to experiment with a similar idea.[3] My idea is to block off a small part of Harvard Square to automobile traffic. This is actually pretty straightforward to do and would likely decrease traffic in the area by encouraging traffic to take faster alternative routes like Broadway or Memorial Drive.

Further, I think it would be a boon to businesses in the area because it would encourage more businesses to use the sidewalks and streets, perhaps setting up more sidewalk seating areas and installing more food trucks or pop-up shops. I think there are a lot of creative ways to enhance the space once more of it is freed up from automobiles.

Whether or not you like this particular idea, I think you can agree that street safety is an important topic in city life and something we, as a city, should be striving to improve.

I hope this post has been informative, interesting, and gets you thinking about ways you can help increase street safety in Cambridge, or whatever city you live in.


[Update: Thanks to an insight provided by a reader, it turns out that the lack of crashes on Memorial Drive may be due to it being a state road, rather than a city road. Which means that only accidents at intersections with city roads would be reported.]


Further Explorations

There is always more to do and learn about. So in that light, a few of the things I thought about while working on this project that I would like to explore further but didn’t include for constraint reasons are included below.

The first is to figure out the denominator, i.e. a better comparison of locations by looking at crash frequency and the amount of traffic a particular location sees each day. As indicated in the post, just because the total number of something is high, doesn’t necessarily indicate a major issue. In this particular dataset, having a high number of crashes in a location as compared to another, may just indicate that there is more traffic in that location, rather than it being an area of concern. What would be more useful would be to look at how locations compare safety wise after factoring in the amount of traffic that flows through it each day.

Unfortunately, this is quite a difficult problem to figure out because it is hard to know how many cars, trucks, buses, people, bikes, scooters, etc. pass through a particular location day in and day out. There is some progress here, like the Bike Counter in Kendall Square, but there still is a lot of data that needs to be captured in order to report on this accurately. Though with advances in machine learning, it could happen faster than anticipated.

The other further exploration would be to look at predictability. Given all this data about the location, objects involved, and time and date, is it possible to predict when and where crashes may happen in the city? Now that there are some really sophisticated machine learning algorithms, it would be interesting to test this out on the dataset. The caveat is that the dataset isn’t huge. But perhaps the way to do it is to use datasets from cities all over the country or world? Might be hard to predict though given different rules, cultures, and driving abilities. Still interesting though!

If you have any interest in solving either of the issues above, please let me know!


Slide Show

I also put together a slide show with all of the data and graphs talked about in the post for easier sharing and for those who’d rather look at pictures.


All of the data used is available from Cambridge’s Open Data portal found here.

For replication, checking my methods, or your own interest, all the code and data I used for generating these findings can be found here.


Enjoy this analysis? It’s part of a series of data analyses I am doing on the city I live in, Cambridge, Massachusetts. You can find the rest of them here: https://medium.com/a-look-at-cambridge.


[1] I took a look at data from the city I live in, Cambridge, Massachusetts, using the city’s police crash data that’s available from the city’s open data portal.

Something you should take into consideration while reading this post and looking at the data is that it is likely there are reporting issues in the data itself. I tried my best to clean it and account for any potential issues, but anything that requires manual reporting in potentially very stressful situations is likely to have inaccuracies and bias in the reporting.

One example of the poor reporting is a number of locations for crashes were reported at City Hall. While there is an intersection near City Hall that could have a number of crashes, the amount reported (by far the most total for a pair of coordinates) seems much too high for this area. I would guess that the reason City Hall has such an overwhelming amount of reported crashes is that it it is used as the default location when nothing else is reported. This is why you may see a large amount of crashes on the map right where City Hall is.

[2] It is also possible that it has nothing to do with the separation of automobiles and bikes. Though I would like that there is a good amount of bicycle traffic on Memorial Drive bike paths, it is possible that this is much smaller than the bicycle traffic on Mass Ave.

[3] If you like this idea and live in Cambridge, be sure to also support it in this year’s Participatory Budgeting. You can do that here.