Random Forest Tutorial: Predicting Crime in San Francisco


Announcement: Annalyzin is now Algobeans! We are also creating a new mailing list that helps us send better emails to you. Click here to update your subscription! This will move you to our new mailing list and remove you from the old one.

Can several wrongs make a right? While it may seem counter-intuitive, this is possible, sometimes even preferable, in designing predictive models for complex problems such as crime prediction.

The Problem

In the film Minority Report, police officers were able to predict and prevent murders before they happened. While current technology is nowhere near, predictive policing has been implemented in some cities to identify locations with high crime. Location-based crime records could be coupled with other data sources, such as income levels of residents, or even the weather, to forecast crime occurrence. In this chapter we build a simple random forest to forecast crime in San…

View original post 1,255 more words