Published August 23, 2010
As promised, here are the first time-series data graphs, the average month by month violent crime index for the entire period for which I have collected/processed data. This bird’s eye view shows a mellow wave that corresponds well to the seasons. Say what you will about Chicago in February, it’s the safest time of year!
Now let’s break this single line down into several based on “side” of the city. Firstly, here is how I defined the “sides”:
And this is the breakdown:
The placement of these lines relative to each other is hardly surprising given the territories they encompass on the crime map. There is a fair degree of arbitrariness is defining geographical “sides” at all, and putting the borders as I have was largely a complete judgment call on my part. Sure, I based it on some level of rationality. I think it’s fair to say that somewhere between 75th St. and 79th St., for example, you’re far enough south to warrant being on the “Far South” side. After all, if you were that far north of Madison, you’d be in Evanston.
As a bit of an aside and a matter of forewarning, I’m going to try to keep this and blog posts generally as free of statistical gibberish as possible to keep this accessible. I’m going to as a matter of policy provide the barest minimum of explanation of the stats in the blog posts. I will hyperlink the concepts to other sites for those who want to read further as well as posting the hard numbers at the bottom of the posts. I am also turning off the moderation settings in the comments, so any real stats geeks who happen to be reading can take me to task for my amateurish methodology.
Back to the graph. It appears from the graph like the high crime areas also have more variation while the low crime areas are relatively constant. This is mostly (but not entirely) an illusion based on the range of numbers the graph is showing. There are certainly larger swings of crime in high crime areas in absolute numbers, but not really as a percentage of total crime. Using the Coefficient of Variation to take this into account, the highest crime South Side has a CoV of about 20.4%, the highest of the bunch. The second highest-crime West Side has the second lowest CoV of about 15.7%. So in actuality, taking into account that there is just more crime in certain places than others, it really doesn’t vary that much more between the various areas mapped here. They all sort of, as we saw in the first graph, roll with the seasons.
|Coefficent of Variation
Published August 22, 2010
In my perpetual quest to walk the line between readability and informativeness, I have divided the colors into groups, roughly modeled after their electromagnetic frequencies, and then created four subdivisions within each color group. I think the added level of complication is worth it. There are the shades of purple on the north, which should come as no surprise. The blues on the northwest and southwest sides almost form a ring around the West Side red zone in Garfield Park. The greens interestingly seem to signal transitionary areas. That is, they are bordered on either side by decent and decidedly dangerous areas. A significant question is whether the crime rate varies uniformly over these areas or changes rapidly somewhere in the middle, with the side bordering on safe areas being itself also safe and the area bordering on the dangerous areas also itself being dangerous. I suspect it is, more often than not, the latter, judging by my various ill-advised bike rides throughout this fair city. I have posted the various VC Indexes below the map. The average for the city is at about 528, which is between the darkest and second darkest green on the map.
And the data:
|Lower West Side
|Near North Side
|Near South Side
|Near West Side
|Greater Grand Crossing
|West Garfield Park
|East Garfield Park
Published April 28, 2010
maps , Uncategorized
Despite the lack of activity on this blog, the Chicago Crime Project has been ongoing. Specifically, I have continued to collect dataand have been improving the method of interpreting the data. As you may have noticed, the data we were dealing with before was all aggregate (i.e the total amount of various crimes in various areas), which is fine as far as it goes (and when we only had six months of data). Now, however, I have two years of data and the time is here to start doing fund and informative things with time series. There will be day to day charts on all manner of issues as we move forward. First, however, is an update of the famous violent crime and property crime maps. They are not only more readable and visually pleasing than the old ones, but I have abandoned the relative ranking I used to do (i.e. the 10 best neighborhoods are one color, the next 10 best another, and so on) and have adopted absolute ranking (i.e. neighborhoods below x crimes per capita are one color, neighborhoods above x but below 2x crimes per capita are another, and so forth). This, I believe, will give a more accurate picture of crime in the city. Note that violent crime has a more fine-grained spectrum than property crime. This is because, for better or worse, there is a much richer and varying fabric of violent crime than property crime in this city. I have also tweaked slightly the formula I use to calculate the Violent Crime Index. More on this later. Now, the maps:
Published December 31, 2008
John and I began researching Chicago Crime and its relationship with demographic factors as students in the Empirical Methods of the Law class taught by Professors Ulen, Robbenolt, and Lawless at the University of Illinois College of Law. The class required a final presentation to the class and a final paper. The presentation can be downloaded by clicking on this link and the paper can be downloaded at this link.
The multivariate regressions in the paper are better thought out than the multivariate regressions in the presentation. The presentation also presents findings are more topics; however, the paper will be easier to follow given its explanations of the findings.
Published December 15, 2008
At long last, here is the promised value-for-the-money map. Essentially, everything low ranked (purple/blues) has low median rent relative to the violent crime. The middle rankings have violent crime rankings well reflected in the median rent, and the reds are high violent crime relative to the median rent.
VIOLENT CRIME/MEDIAN RENT RELATIVE RANK MAP:
And here’s the corresponding table. Those who are observant may notice that the “Value for $” figure is nothing more than the median rent rank minus the violent crime rank. I tried many fancier methods, but this seems like the most crisp and effective way of determining how much daylight there is between the rent and violent crime in a neighborhood. A discussion of the results follows:
Obviously, there are factors that we would expect to make rents skew low. One would be distance from the Loop. Thus, the high value areas on the far Northwest and Southeast sides of the city make sense and are expected. The commute into the Loop in these areas is longer than the commute from many inner western suburbs. What is remarkable is the huge sea of purple on the Southwest side. Even the South Loop, a stone’s throw away from downtown Chicago, is a remarkably good deal. So, what are these neighborhoods, and what do they look like?
||Household Med. Inc.
||Lower West Side
||Near South Side
Quite a diverse lot.
Published December 14, 2008
Here is a map of the median rent (2005) estimations for each of the 77 Chicago Community areas. The colors are a bit different than usual. They are arranged with the lower ranked areas having higher rents. This is a bit odd, but it does make it easier to compare to the crime maps. There are clear similarities between the crime and rent maps, which probably should not come as any surprise.
MEDIAN RENT MAP:
And here is the underlying data in order (highest to lowest rent), now with nifty hyperlinks:
Published December 4, 2008
Thanks first to our small but helpful commentariat. Just because your suggestions have not been acted upon does not mean they are (entirely) without merit. We have been busy actually updating data and running regressions. Expect more substantive posts soon, but, in the meantime, please enjoy an updated version of the previously posted violent crime map as well as a similar one for property crimes. You will note that I replaced the green color with tan to make it more intuitive (and visually pleasing). Again, lower ranked areas have less crime.
Update (12/14/2008): Minor corrections made to violent crime map. 7 Community Areas move up or down one category from previous map.