Level: Beginner
Sherisa Yocher, Graduate Student, University of Connecticut
Saraswathi Sathees, Graduate Student, University of Connecticut
Rebecca Gill, Graduate Student, University of Connecticut
In Malaysia, PetFinder.my has been the number one animal welfare platform since 2008. For this analysis, more than 14,500 pet profiles have been analyzed to find key insights that will help shelters and other non-profit organizations craft future pet profiles that will lead to the fastest adoption rates possible. We seek to reduce the amount of animal suffering and euthanasia while bringing together happy pet families.
The sample data was supplied by PetFinder.my and compiled from a set of five original data files. The initial data set contained 18,481 rows and 28 columns of data. After the data was pre-processed, the data set contained 14,991 rows and 15 columns of data. The target variable was adoption speed (within 30 days or not; Yes = 1, No= 0) which was predicted using 14 statistically significant variables.
Seven models were created to predict pet adoption speed, including Decision Tree, Bootstrap Forest, Boosted Tree, Neural Network, K Nearest Neighbor, Ordinal Regression and Naive Bayes. After assessing all seven models based on the misclassification rate, RMSE, accuracy of 1’s and total accuracy, it was determined that Naive Bayes was the best model. Based on the results of the Naive Bayes model, key recommendations for PetFinder.my include the following: First, each pet profile should feature only one pet, rather than a litter or grouping of multiple pets; second, each pet profile must include at least one photograph and not exceed 10 photographs; third, PetFinder.my should consider transporting pets from low adoption rate states to other states with higher adoption rates; and lastly, unhealthy pets should be nursed back to health before being posted for adoption, despite associated fees. Implementing these key recommendations should decrease the adoption speed of homeless pets and reduce the amount of animal suffering while bringing together happy pet families.