Author: Matt Sandy

Flatten Data in R by Column Values

Not sure if this will help anyone, but I recently wrote a script that will let you flatten/widen a data set based on a dynamically set column name.

Animal Instances Instances Instances Instances
Puppy 47 77 97 25
Puppy 26 6 14 98
Kitten 38 50 5 73
Puppy 83 61 43 4
Kitten 78 30 57 8

…would turn into this

V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12
Puppy 47 77 97 25 26 6 14 98 83 61 43 4
Kitten 38 50 5 73 78 30 57 8

Using Police Data to Track Citizens

While looking for city data to keep up with my data analysis and visualization skills I found a disturbing data set.

The Minneapolis Police Department uses automatic license plate recognition to help aid in solving crimes in real-time. They also maintain this data about innocent citizens for months after it was collected. These data were obtained legally, via the Minnesota Government Data Practices Act. They have been pre-processed and deidentified. This file contains over 800k plate readings. – github.com/johnschrom/Minneapolis-ALPR-Data

I decided to check it out, and see what a limited exploratory analysis could determine from the top 20 license plate hits, I will start with the biggest issue (in my opinion),

Determine When a Person is Home

Here is a map of the top 20 license plates and the data points for when the police scanner ran their plate.

License Plates Aggregated
License Plates Aggregated

Now if we look at a time where people are most likely at work (say 11am) we can grab the hashes of the license plates that work at specific locations.

License Plates at 11am
License Plates at 11am

You can see above that the purple person is almost always SE of the cities, The lat/long is 44.935696,-93.38466 which corresponds to the Home Choice Parking lot in Hopkins, MN. Where will they be 12 hours from then?

License Plates at 11pm
License Plates at 11pm

Above you can see they pretty much always return to Dinkytown, Minneapolis. Using this method you could potentially

  • Determine whether or not a person is likely to be home at a certain time
  • Determine where someone likely lives
  • Determine where someone likely works
  • Determine if your employee took their car out when they called in sick (two data points already known)
  • Determine what someone is up to on their time off (if work and home locations known)

Obviously with more data points the more reliable the data becomes, and this was only run for a few months with much of the data tagged with a 0/0 for longitude and latitude.

Using Police Data to Track Targeted Neighborhoods

Using the same data it is also possible to track which neighborhoods are targeted by these scans

Plate Counts
Plate Counts

Brunch At The Farm

The Farm @South Mountain is an incredibly cute spot. I recommend anyone in Phoenix going there at some point, but make sure to look up their hours before.

The Colors of the Arizona Desert

Took this January 1st (along with a bunch more photos) while traveling in a hot air balloon with my girlfriend and friends Ash, Katrina and Michelle. It was an amazing experience, and we went through the company Float which is run by a great couple. I will be posting more photos from the trip once I scan the rest of them.

Phoenix Flea Market

Some guy looking pretty cool
Man in bowtie
Woman flipping through some records
Another mobile thrift shop
Peace

Phoenix Flea Market