Finding Forever Homes: How to Increase Cats’ Chances of Adoption from Shelters
As we head into the spring and summer months, “cat season” is upon us, meaning that animal shelters and sanctuaries across the globe will see an influx of intakes as more kittens are born. Apart from providing a secure place for stray animals, the primary goal of most shelters is to get their animals adopted by families in individuals, in order to keep making room for newer ones. But shelters often struggle with limited resources and funding, limiting the number of animals they can care for at one time and making it more difficult for them to get their animals adopted.
With this project, I examine data on shelter cat adoptions in Malaysia (provided by PetFinder.my as part of their 2019 Adoption Prediction code competition hosted by Kaggle) with the intention of finding which features show the strongest association with quicker adoption speeds (i.e. less time spent in the shelter) and, ultimately, making recommendations to shelters on how best to utilize their resources, as well as how to advertise and promote their adoptable cats, in order to get them adopted as quickly as possible, and matched with a forever home.
Several of the features in the dataset are indicated categorically and need interpretation. Explanations of the categories will be explained where relevant, but the primary outcome being examined is adoption speed, which is indicated in the dataset as follows:
- 0 — cat was adopted on the same day it was listed
- 1 — cat was adopted 1–7 days after being listed
- 2 — cat was adopted 8–30 days after being listed
- 3 — cat was adopted 31–90 days after being listed
- 4 — cat wasn’t adopted after 100 days
So in general, lower number of adoption speed indicates a faster adoption after the cat became available.
The dataset included many detailed descriptors of the cats and their adoption profiles, but as preliminary data exploration, I examined the correlations between some of the more quantifiable/binary features included. The features I examined were gender, age, fur length, size, whether the cat has been sterilized, vaccinated, and dewormed, the cat’s health (specifically relating to physical injuries), the adoption fee, and the number of photos and videos posted in the cats’ adoption profiles.
Somewhat unsurprisingly, the strongest positive correlations appear between the binary variables indicating the cats’ vaccination, sterilization, and dewormed status. These are considered some of the most vital check-ups to have completed before bringing a cat home (especially to a home with other pets), and when bringing in a stray cat who has spent most or all of their life outdoors, most vets will typically perform all of them at the same time or in close succession, to ensure that they don’t spread diseases and worms to other cats, or have more kittens.
Another positive correlation of note appears between fur length and adoption fee, suggesting that cats with longer fur, which are typically closer to being full breeds and also seen as more luxurious, draw higher revenue for the shelter when they are adopted. While most shelters aren’t selective about the types of cats they take in, this does suggest that it may be beneficial for them to emphasize longer-furred cats (when they have them) in advertising adoption profiles.
Given that many people would generally be unwilling to take a cat into their home if they knew the cat would be able to have kittens with currently owned cats or other cats in the area, the first features I want to do a closer examination of are health features, specifically of whether a cat being sterilized (i.e. spayed or neutered) and whether a cat had physical injuries would affect the speed of adoption.
For sterilization, 1 indicates that the cat has been sterilized, 2 indicates that it hasn’t, and 3 indicates uncertainty. For health, 1 indicates no injuries, 2 indicates minor injuries, and 3 indicates serious injuries. The lines on the violin plots are indicative of quartiles.
Interestingly, it appears that cats who haven’t been sterilized are adopted much faster than those who have, and slightly faster than those whose status is uncertain. However, keeping in mind the correlation plot above, there’s a relatively simple possible explanation for this: the younger the cat, the less likely it is they’ve been sterilized, since the optimal age for sterilizing kittens is between 4–6 months. And, unsurprisingly, young cats, particularly kittens, are in higher demand for adoption than older cats. So this would indicate that when people are intent on adopting kittens, whether or not the kitten has been sterilized doesn’t play a role in their decision.
However, cats’ physical health does appear to affect the speed at which they’re adopted: Cats with no injuries get adopted the quickest, with cats with major injuries waiting around median times between 1–3 months. It’s reasonable to assume that cats currently being treated for severe physical injuries would not be listed for adoption until they were as recovered as possible, and given that adoption speed is calculated based upon when the cat was made available for adoption, and not taken in by the shelter, it would appear that major physical injuries (such as missing limbs, ears, or eyes) deter people from adopting cats.
Given this, it could then be beneficial for shelters to even more strongly emphasize these cats’ positive traits (affectionate, calm, etc.) when listing them for adoption. More on textual analysis of the listing descriptions will be discussed later.
Adoption fee is another important factor to look at in regard to adoption speed, given that most people view shelters as a lower-cost alternative to breeders and pet shops. However, shelters still need financial support in order to care for their animals, so it can be hard to find the right balance between the two.
The data still shows some semblance of discreteness despite being on a continuous scale, since many shelters likely have different standards and bases for how to determine adoption fees. Slower adoption speeds to show stronger concentrations of higher fees, indicating that shelters that set their fees too high may have a harder time getting their cats adopted. However, pets that were adopted within the first month (adoption speeds 2 and 3) also show concentrations around non-zero fee levels, so fees between 50–200 ringgit (equivalent roughly to $10–50 USD) appear to be a so-called “sweet spot” for ensuring shelters can stay afloat without overcharging or losing out on revenue.
And as one might expect, many of the cats that were adopted the same day of listing had no adoption fees, so if a shelter is having trouble getting certain cats adopted (as in the above example with health), lowering or eliminating the fee could be a viable way to get them adopted.
Earlier, it was mentioned how kittens are in higher demand than older cats, and thus get adopted quicker. While this observation may seem anecdotally true, there is evidence to support the notion that younger cats spend less time in shelters.
The strongest concentration is around 18 months, which is widely considered the cutoff in maturity between kittens/young cats and adults. Many individuals and families are looking for kittens (as indicated by the still-strong concentration for less than 18 months), but many are unable to care for a full kitten (whether do to presence of other pets in the house or inability to provide the full amount of care/attention needed), while still wanting a young cat that they will get to spend more of its life with. While 18 months is the center of the concentration, adoption speed does get slightly quicker between 2–8 years of age, before dropping off. This could once again mean that shelters need to emphasize the positive traits of older cats (greater than 8–10 years of age) when listing them for adoption.
Crafting a good post on the shelter’s website is essential to raising awareness of the cats and raising their chances of getting adopted. One of the key components to these posts are the pictures and videos posted of the cats, which let potential adopters know the cats’ appearance and personality.
While having more pictures and video is slightly associated with faster adoption speeds, having more than 10 total pieces of media doesn’t appear to drastically increase the speed at which the cat is adopted. Having no pictures and video listed with the cat’s post, however, drastically raises the amount of time spent in the shelter (nearly all of the data points for 0 photos/videos are in the adoption speed 4 category), indicating a clear need to include at least some form of media in the post.
Finally, keeping in mind all of the previous insights gained from looking at the numeric data, I examined the text of the descriptions that accompanied the cats’ listings. In doing so, I looked at the words in posts that were most strongly and positively correlated with the fastest adoption speeds.
Somewhat contrasting with an earlier finding, “spayed” appears as a strongly correlated word. The findings above took into account sterilization across gender (meaning spaying and neutering combined), but this could mean that female cats being sterilized leads to them being adopted quicker. This would make sense, given that a pregnant female cat is much more trouble to deal with (due to the health risks and future kittens) than a male cat who has fathered kittens. It could then be worthwhile for shelters to focus upon spaying their female cats, rather than splitting efforts/funds evenly between male and female cats.
“Whatsapp,” a popular messaging app, also appears, potentially indicating that the shelter providing an easy line of communication with potential adopters is an important factor in getting cats adopted quickly.
And, as suggested before, words like “love,” “friendly,” “sweet,” and “best” all appear, which would indicate that descriptions that emphasize cats’ positive traits, rather than focusing upon the circumstances in which the cat was found (as many descriptions in the dataset did) show more success at getting cats adopted. It then would be worthwhile for shelters to put more of their effort (accomplishable by volunteer behavior, rather than needing funding) into crafting detailed posts advertising their cats, so that potential adopters are as informed as possible, and more willing to take the plunge and give a cat its forever home.
Sarah Potts is a junior in the College of Arts & Sciences at the University of Pennsylvania studying mathematics, data science, & creative writing. This data project was created for Prasanna Tambe’s course Analytics & the Digital Economy (OIDD 245).