Animal Shelter Data | Data by Jamie

Louisville, KY Animal Shelter Intakes

My analysis uses data found on data.gov, published by Louisville Metro Government. It's the analysis of the intakes of dogs to their local animal shelter from January 1, 2020 to April 6, 2024.

Problem

Why are dogs entering animal shelters?
Can illegal backyard breeders be located using data?

Background

Ever since I adopted my Mini Australian Shepherd, who was likely a result of backyard breeding, I've been extremely interested in the issue. For context, backyard breeding is the act of illegally and unprofessionally breeding animals to sell. The more I work with animal rescues, the more I realize how overcrowded shelters are and how most of the dogs in shelters were just "found" with no way of knowing where they came from. Shelters and rescues are only focused on getting dogs adopted because they need to make room for more animals that keep coming in. Most organizations don't have time to look at how to stop the problem. This motivated me to look into what's happening to these dogs before they enter the shelter, and how much backyard breeding is contributing to the issue.

One issue that makes it really hard to find patterns is the lack of good and uniform data from animal shelters. Because of this, I decided to start my research using data from only one shelter. Louisville Metro Government publishes a really clean and complete dataset relating to the intake and outcomes of animals which enter their shelter.

Solution

The dogs that enter the shelter are 83% strays, 7% surrendered by the owner, 5% returned to the shelter after being adopted, and 4% confiscated.
Using this dataset, I cannot say for certain where there may be illegal breeding, but one area that could be investigated outside of data analytics is Buechel, KY (zip code 40218). In Buechel, 1,976 stray dogs have been found over the last 4 years. 571 of those dogs are Pit Bulls, which is at least 4x higher than Pit Bulls found in any other area.

Approach

To focus my analysis, I'm going to look for trends in breed by location, spay/neuter trends, and pure bred vs. mutts.

This dataset was very clean. Most values were already standardized and all data was capitalized. The only cleaning I had to do was reformatting the dates and datatypes. From there I filtered the data to only show dogs that need homes (as opposed to dogs that were lost/found and returned to their owners).

Once I filtered out the data which was unnecessary for this particular analysis, I wanted to know why these dogs were taken to the shelter. So I created a pie chart to also see how many dogs were taken in for each reason. I found that majority of the dogs were strays, which confirmed what I had learned working with animal rescues.

Since 83% of the dogs that enter the shelter are strays, I decided to filter the data further and focus on stray dogs only. The data provides a reason why the animal was surrendered, so I created another pie chart to see why the stray dogs were surrendered. Unfortunately, I found that 96% of those strays don't have a reason for being a stray. For context, the reasons the other 4% of strays were surrendered included being abandoned, the animal was aggressive, the owner died, etc.

From there I decided to see how many of these strays are neutered, spayed or intact. I expected most of these dogs to be intact, as there are so many strays that I thought might continuously be reproducing. I was also hoping to have a larger pool of intact dogs to analyze when searching for backyard breeders.

I created two categories, fixed and intact. To my surprise, 81% of stray dogs are fixed, meaning at one point these animals were all likely pets that were given up.

I built a Tableau heat map to locate all the stray dogs to see if there are any patterns. The map shows the density of stray dogs found by each zip code in and around Louisville, KY. The first thing I noticed was that the zip code 40218 looked significantly darker than any other area. I went back to RStudio to find that 1,976 stray dogs were found in zip code 40218, with the runner up being zip code 40216 with 839 stray dogs. I used the dashboard to filter out the dogs that have been fixed to narrow down my search for breeders that might be abandoning dogs that aren't selling. Then I filtered for dogs that appear by the data to be pure bred, though it's important to note that you can't know a dog's breed for certain by just looking at it, so the data on breed is a best guess. Then I filtered for specific breeds to see if anything stood out, but no matter how I filtered the data the trend was more or less the same, which is that the count in zip code 40218 is normally always the highest.

I decided to look deeper into the breeds in 40218, which is Buechel, KY. Although the total stray dog count in Buechel is at least 2x more than any other zip code, the count for Pit Bulls is at least 4x higher than any other zip code. Most of these Pit Bulls are fixed, but that doesn't necessarily mean backyard breeding isn't involved.

zipcode breed n
<chr> <chr> <int>
1 40218 PIT BULL 571
2 40211 PIT BULL 164
3 40216 PIT BULL 162
4 40212 PIT BULL 150
5 40208 PIT BULL 123
6 40214 PIT BULL 122
7 40215 PIT BULL 120
8 40218 PIT BULL / MIX 120
9 40219 PIT BULL 94
10 40218 GERM SHEPHERD 92
# ℹ 4,436 more rows

Feel free to view my code in GitHub!

Conclusion

Most of the shelter dogs are strays and Pit Bulls. This is no surprise as this is the typical trend nationwide. It's also common knowledge that illegal backyard breeding contributes to the large number of Pit Bulls that are abandoned, but when you can narrow it down to a single zip code that has significantly higher numbers than the rest, that's a great place to perform further investigation.

It's important to note that Buechel, where the highest number of stray dogs were found, is also where Louisville Metro Animal Services is. There could be a few reasons for this:

Simply coincidence
Dog owners drop their dogs off in the area knowing that the shelter is there
The shelter inputs their own zip code when they don't have the correct data

These are points that would need to be discussed with the shelter and reevaluated.

It's difficult to track down the origins of stray dogs, especially when shelters, unlike Louisville Metro Animal Services, have inconsistent, incomplete, and unstructured data. Hopefully with more standardization to how shelters collect data (as Shelter Animals Count is working on), and with more people paying attention to where animals are coming from, we can more easily find breeding grounds and slow the rate at which dogs enter animal shelters.