Comparison between the canine behavioral assessment and research questionnaire and monash canine personality questionnaire - Revised to predict training outcome in apprentice assistance dogs

Ratings of canine behaviour and personality are a convenient method used by dog training organisations to gather information about prospective and trainee dogs. The objective of this study was to compare the use of two rating questionnaires to predict training outcomes in assistance dogs. It was of interest to investigate the predictive power of a questionnaire answered by the dog trainer in case no puppy raiser questionnaire was available. Two standardised ratings were used, in particular, the Canine Behavioral Assessment and Research Questionnaire (C-BARQ) was completed by puppy raisers around the time the dogs started formal training and the Monash Canine Personality Questionnaire - Revised (MCPQ-R) was completed by dog trainers at ten weeks of training. Rating data were independently analysed to investigate their relationship with training outcomes. The results from the univariate logistic regression analysis were used to select the variables for the reduced feature sets that were used for modelling. The novel machine learning models built with data collected using the C-BARQ and MCPQ-R achieved similar performance in predicting training outcomes, an area under the ROC curve of 0.84 and 0.85, respectively. The novel models developed in this research were the most effective early prediction of suitability for assistance work compared to previously reported studies. The MCPQ-R was demonstrated for the first time to be a reliable canine behavioural assessment method for estimating future outcomes in trainee dogs. The dataset and code used are publicly available on GitHub.