Table 3. Search results among all the models of non-parametric tests of Friedman and paired comparison of all models.
A comparison was made between all models with the non-parametric Friedman test and a pairwise comparison of all models. The best predictive capabilities in the dataset were shown by the Random Forest approach. In addition, it must be noted that such signs as Muscle Thickness, Back Fat, Average Daily Gain can act as predictors of leg weakness. Information on breed and gender were not significant for assessment the status of legs.
| Model A | Model B | p-value | Model A | Model B | p-value |
|---|---|---|---|---|---|
| boosting | arbol | 4.37E−02 | NB | logistic | 2.17E−08 |
| KNN | arbol | 2.17E−08 | NET | arbol | 2.17E−08 |
| KNN | boosting | 2.17E−08 | NET | boosting | 2.17E−08 |
| LDA | arbol | 2.17E−08 | NET | KNN | 2.17E−08 |
| LDA | boosting | 2.17E−08 | NET | LDA | 1.31E−07 |
| LDA | KNN | 2.17E−08 | NET | logistic | 1.31E−07 |
| logistic | arbol | 2.17E−08 | NET | NB | 2.17E−08 |
| logistic | boosting | 2.17E−08 | rf | arbol | 2.17E−08 |
| logistic | KNN | 2.17E−08 | rf | boosting | 2.17E−08 |
| logistic | LDA | 9.30E−02 | rf | KNN | 2.17E−08 |
| NB | arbol | 2.17E−08 | rf | LDA | 2.17E−08 |
| NB | boosting | 2.17E−08 | rf | logistic | 2.17E−08 |
| NB | KNN | 2.17E−08 | rf | NB | 2.17E−08 |
| NB | LDA | 2.17E−08 | rf | NET | 2.17E−08 |
| Friedman rank sum test | |||||
| Friedman chi-squared = 286.85, df = 6, p-value < 2.2e−16 | |||||