Table 1 Model comparison.

From: Predicting fertility from sperm motility landscapes

Model

Boar

Sow parity

Ejaculate

Motility

ELPD difference

SE

M0

x

-

-

-

−2.8

3.0

M1

x

x

-

-

−1.9

1.9

M2

x

x

x

-

−2.6

1.9

M3

x

x

-

x

0

0

  1. A series of models were constructed as described in methods, "Fertility modelling" section. Null models (M0, M1, and M2, without motility information) were compared to models based on different landscape configurations, M3 being selected with an 11-cluster configuration, using an approximation of leave-one-out cross-validation and relative predictive performances (ELPDs) as scoring metrics (ELPD = 0 corresponds to the highest predictive ability). The model parameters are: Boar = inter-boar variability (as random effect), Sow parity = number of cycles of pregnancy of the sow, Ejaculate = intra-boar variability (as random effect), Motility = proportions of sperm in motility clusters, and the proportion of static spermatozoa of each boar. The last column (SE) shows the standard error of the ELPD difference, relative to the model with the highest ELPD (M3). See details about ELPD in loo R package documentation and Vehtari et al, 201779).
Back to article page