If you love dogs, you will love NYC. There are many dogs in NYC; you will always spot someone walking their dogs on the street. New York has over an estimated 1 million pets, while about 60% are dogs. Dogs or any pets are becoming the new “child” of the house. According to City Journal, millennials are most enthusiastic about being a pet owner, with some 70 percent boasting of having at least one pet. In some cities such as San Francisco, there are more pets than children. Even Though pet owners look after their dogs and provide the best training possible, there are always incidents that are very unfortunate to both humans and our beloved dog friends. In 2018, 39 people died as a result of dog bites. There are an average of 6,000 emergency department visits related to dog bites each year. Dog bites are a leading cause of nonfatal injuries to children. Lastly, Most dog bites are from dogs who have an owner. In this report, I will be focusing on Dog Bite Incidents in NYC from 2015 to 2017.
Step 1: Gather Data
I downloaded CSV files from the NYC Open Data. There are many valuable data on various topics, and I chose Dog Bite data since that is the issue I would like to explore the most. I decided to work with two datasets related to Dog incidents, such as Dog Bite in NYC and Dogs in NYC. It is necessary to look at both data to get a broader idea of NYC dogs’ situation.
Step 2 Data Preparation: Google Sheet & Open Refine
The data provided was in a usable format(CSV) but uploaded to Google Sheet because it is the most convenient way to work and adjust later. I went through the data and made sure it’s cleaned, and I deleted several columns that are unnecessary for this report and reduce the size of the file. I also used Open Refine to edit the redundant data and edit the exiting Google Sheet file to update the date in Tableau without re-uploading the new file.
Step 3 Visualization: Tableau
I linked my updated Google Sheet directly to Tableau Public to make the datasheets and dashboards. This software/platform is helpful for both visualizing data and presenting the analysis. I created two dashboards, one for dog bite incidents and another for the number of dogs in NYC, to see if there’s any relation between the two. To keep data consistent, I only analyzed data from 2015-2017.
Results and Findings
Increased dog incident when temperature gets warmer
The above chart is about the number of dog bite incidents between 2015-2017 in New York City. It seems like every quarter three; there’s a peak of incidents. I assumed it’s because the weather is warmer, and people more likely to spend more time outdoors, including their dogs. In the winter season, people tend to stay indoors, so the number declined. I also looked at the number of dogs registered around the same time. There is also an increase in dog registrations in quarter three. There might be a possibility that puppies or young dogs are accidentally biting people, or it might be dogs are not used to their new environment(see chart below).
Breed with most incident cases
The breeds with the highest risk of the incident are Pit Bulls and American Pit Bull/Pit Bull(mix)s. They are not the most popular breed in NYC but have the highest incident rate. According to Forbes, Pit Bull is responsible for the most fatal attacks in the U.S. The following infographic shows that the Pit Bull is still accountable for a large number of fatal attacks in the U.S. by far, killing 284 people over 13 years or 66 percent of total fatalities. That’s despite the breed accounting for just 6.5% of the total U.S. dog population.
According to our data, other than Pit Bulls, Shih Tzu and Chihuahua are among the 2nd and 3rd “dangerous” dog breeds and ranked top 2 and 3 most popular breeds in New York City. There are about 1,587 dogs’ breed are unknown, so that might change the analysis.
Map & Borough
Other than showing the breakdown by Borough like the graph above. I generated maps using multi-colors with hierarchy and charts for boroughs to show viewers where incidents happened from high to low. The viewer can also use map filters to detail specific locations, such as zip code 10029, which has the most cases from 2015-2017. Zip Code 10029 is right next to Central Park where many people walk their dogs.
Boroughs that have the most number of dog bite incidents are Queens(2,520 cases), Manhattan(2,354 cases), and Brooklyn(2,281 cases). I also attached a map that shows the total amount of dog bites versus dog registrations in NYC so viewers can compare. The area with the most incidents seems like they have more dogs.
Young dogs are more likely to case incidents
Younger dogs have a higher rate of biting someone compared to older dogs. Dogs with a higher risk of attacking humans are from age 1 to 3. Age 2 had 504 cases, age 1 had 438, and age 3 had 424, then the number started to drop increasingly to 324 from age 4.
Spayed or neutered dogs are less likely to bite someone
Not spay or neuter dogs are more likely to be aggressive to attack someone. According to the VCA, the castration of male dogs reduces or eliminates some forms of aggression because male hormones influence. Some debate on whether or not spay or neuter will help with aggression behaviors, but our data show that the spayed/neutered dogs caused fewer incidents than those who are not.
Male dogs have higher chase to cause incident than female dogs
From the known data we have, male dogs are more likely to cause incidents compare to female dogs; with known data, about 2/3 of the dog bites are by male dogs.
I conducted two user testing to make sure my visualization to be easy and clear to comprehend, one from the UX field and another from a non-information-related field. The method will be Zoom interviews using the think-aloud method. My goals:
- Explore the dashboard for several minutes and let me know your thoughts.
- Ask questions regarding the data to see if the participants can discover the answers.
- What is the relationship between the two graphs? Does data coordinate with one another?
- Is there anything confusing?
- Can you tell which dog breeds or locations have a higher risk of accidents?
- Is there any trend you see in these graphs?
User Test Findings
By listing breeds, gender, age, spay/neuter condition in each area, the reader could easily find out which breed group is more likely to have aggressive behavior. I provided a couple of hints on working with the dashboards, but overall it was easy for participants to go through the dashboards to understand data. I used two colors for the maps to distinguish the dataset easier and make the content on both dashboards layout in a similar format to limit confusion.
- The title helps understand the content of the visualization
- Users enjoy the flexibility of changing dates, filter specific data, hover for more information.
- Isolated incidents to see what is going on and Giving the researchers the flexibility is nice
- Unable to figure out why certain areas have a higher risk of being bit by a dog, for example, zip code10029.
- Curious about how human behavior could influence bite incidence.
Adding data of neighborhood population and reason that trigger dogs to be aggressive. Also, if the dog is leashed or unleashed when the attack occurs, maybe that could be another variable statistic that could affect the bite number.
Associate some data regarding human behavior, which would be helpful to analyze the reason behind the attack. Humans are larger than most dogs, so maybe human behavior impacts dog more than the environment or the breed itself. Simply blaming the breed, age, and other conditions may not be fair. If could provide more side study/information about the impact of human behavior related to the dog attack incidents, it could be more helpful to avoid the bite.
In addition to the above, maybe the human behavior study could also explain why the particular area has more dog bite incidence than other areas due to the residential density.
I am happy about the overall turnout of the visualization. Even Though I felt like there should be more information added to make accurate assumptions on the data, I created a cohesive visualization with all the information I have on hand. One of the limitations is too many unknowns in each category, such as location, genders, spay/neuter, or breeds.
If I took this project further, I would like to look into human behavior associated with dog aggression. Also, I would like to explore the source of the dogs, for example, whether they came from reputational breeders, rescue centers, puppy mills, or others. I believed it could significantly impact dogs’ behavior and temper.