Data Science Capstone — Crime and Real Estate, part 2

Scott Biggs
Jul 10, 2017 · 2 min read

I have been able to get all of the data that I need. The last hurdle is to create the models and run it through it. I have split the data into a train test split of 80/20, with the 20% being data from around 2015 to current. I did this because I believe crime is already random and does not need to be randomized further. If the model fits well with both datasets it should also fit with any new crime that occurs, at least in the area where the model was created, logically anyways. Another test of the m0del, and to prove my point further I am also going to run the crime through the model in a yearly format to see if the accuracies hold true.

All of the above being said I decided to added the training data to Tableau and noticed something interesting. The dots where the crimes mostly occurred were near main streets, and more frequently still when those main streets were near highway access.

Sandy Springs Crime Data 2009 to May 2015 / NonViolet Crime (Blue) , Violent Crime (Red)

I was able to add the streets layer into Tableau after noticing this and I was able to confirm my assumption. Still standing by my original hypothesis that house value may play a role in the crime, I also had a new factor to look into.

My biggest hurdle at this point is getting KNN to work with the latitude/longitude data I have for the crime data along with my house pricing and apartment data to add the layer of needed correlation. Once I have this piece of the puzzle the rest of it should fall into place, with the exception of the new found discovery of the main road proximity. This may end up being a reach item, but I want to find out the correlation of proximity to a main road and the rate of crime. I am sure this task will prove to be a much larger under taking that the house pricing relationship and will require further tool sets which I am eager to dig into.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade