Crime vs. Temperature

More than 5,000 publicly available crime reports were processed (from 2001 to that day) and when compared with past weather data, we found that this traditional wisdom is often eliminated.

Like many other citizen hackers in the city, I am pleased to see the recent release of the Open Data API of the Chicago Police Station. After several weeks of hacking, I was very pleased to announce that I tried very basic visual effects.

The number of crimes committed at a specific temperature divided by the average number of days

To do this, I downloaded a complete CSV dump of over 5,000,000 crimes reported in Chicago in 2001, then collected weather data from 2001 for NOAA and loaded it into MongoDB I started watching it. What I can see Technically there are not too many obstacles to overcome. MongoDB comes with a convenient Playboy method for importing csv data directly into the database, so it's a bit patient (it takes about 5 hours to import criminal data) and you can easily load everything. I created several indexes to make things faster and for criminal data, I have created a "location" field containing GeoJSON objects.

But, in fact, it's all about getting this on the web. This is to do this with very few (ie, no) resources. You do not need to worry about the type of managed backend configuration. I need to stay on my local machine. With this in mind, the site's persistence layer (including map content) is all supported in flat JSON files. In this way you can run the application completely from S3 bucket without server side. I do not know how much I can push, but now it works. There are some obvious disadvantages such as not being able to query the contents during flight, but for the moment I only want something that provides a view of the fire hose.

My goal is to update this website daily when new data crime occurs. I hope to be able to automate more, but for the moment I just wrote a few scripts to handle heavy work, but I need to trigger it every day. Another unfortunate thing is that MongoDB is on the VM of my laptop. I would like to share it with someone and come up with an easy way to support it. It's not too expensive (although you can breathe it and put it in S3 or some other thing).

Do you think that you like chicken? Okay, you can ask me anytime. This is feasible. If it does not work, you can apply an algorithm similar to Netflix's behavior. Rather than drama, crime, comedy, or movies from the 1970's to the 1980's, the uses are categorized according to the application (protein, taste, texture), then apply group knowledge to previously cooked recipes ( This is called collaborative filtering.) This type of machine learning can be used to search for recipes. Do you think I want to eat chicken tonight? It is one thing to figure out if you like chicken. It is another thing to wonder if you want to do it tonight. This is true, because in addition to the internal food stimulus (taste and preference), food decision can be made based on other factors. These factors include social environment (meal only, or eating with friends), location, conceptual factors, and anticipated impact (you can eat more tonight as you will be hungry tomorrow) Yes.

A murder investigation outdoors may be a catastrophe of regulated environmental crime scene. This is because when investigators try to reconstruct crime scenes, countless factors can be dazzled from unstable temperatures to the number of people passing through that area. When doing this, it is important to try to find all the will, it is difficult for a long time, in that case it already exists. Forensic anthropologist - one of the top ten people in the world - involved, but he can not give them to him any time he is dead because he is dead