Domestic hens lay eggs in sequences of one or more eggs, separated by one or more non-laying days called a pause. Because of variability in the maturation process modulated by hormone and environmental factors, the exact day and time of oviposition events during the laying period is highly variable. Given that oviposition contributes to nutrient requirement, identifying oviposition events might be helpful in future precision feeding applications aimed at increasing feed efficiency and reducing excretion of N, P, and CO2 to the environment. In a flock of free run birds, it is difficult to determine the daily oviposition event of each individual bird. The objective of the current research was to develop a random forest classification model to predict oviposition events of free run broiler breeders fed by a precision feeding system, based on observations recorded by the system.
A total of 202 Ross 708 free run broiler breeder pullets were randomly allocated to 8 environmentally controlled rooms. Each room was equipped with a precision feeding station. Birds were trained to use the precision feeding station After 17 d, if a bird’s BW was lower than the target BW, the bird could have a meal and then be ejected. Birds were immediately ejected by the precision feeding station without any provision of feed if their BW were greater than the target BW. Daily oviposition events were confirmed by palpation every morning at 7:00 AM. A total of 2,364,761 observations in the station dataset were processed to extract information such as age, BW, feed intake (FI), the time interval between 2 meals, number of machine visits, and successful meals per day. Important features were selected using random forest-recursive feature elimination with 5-fold cross-validation, and 28 features were selected to build a random forest classification model.
The highest accuracy was 0.8522 when the first 28 features were selected. Among the 28 features, FI was the most important one. Precision (a measure of purity in retrieving) of no egg-laying and egg-laying, recall (a measure of completeness in retrieving) of no egg-laying and egg-laying were 0.8814, 0.8090, 0.8520 and 0.8453, respectively. The Kappa coefficient of the model was 0.6931, indicating substantial agreement (substantial agreement range: 0.61–0.80).
To our knowledge, this is the first time the random forest algorithm was applied to predict daily oviposition events in broiler breeders. The dataset from the precision feeding system was able to predict the daily oviposition event of free run broiler breeders with an overall accuracy of 0.8482. Since oviposition is a contributor to nutrient requirements, identifying egg-laying events of free run broiler breeders would be beneficial for providing the birds with precise amounts of feeds appropriate for individual egg production levels, which would increase feed efficiency, reduce production costs, and reduce excretion of N, P, and CO2 into the environment. To improve the prediction accuracy of the model, the biological characteristics of broiler breeders and other machine learning methods might be considered.