Figure 7 | Scientific Reports

Figure 7

From: Statistical model choice including variable selection based on variable importance: A relevant way for biomarkers selection to predict meat tenderness

Figure 7

Parallel boxplots of MSEtest made on 100 learning/test replications (A); repartition (in %) of the number of selected biomarkers in each model (B); and percentage of occurrence of each biomarker in the selected model (C). (A) The more the MSE is weak and the less the boxplot is displayed, better are the results. (B) The model might be considered as stable if the percentage of a given size of the reduced model is significantly higher than the other. (C) The most often a variable is selected in the model, the most important is this variable as predictor. Nevertheless, if a high number of variables appears to be selected in the model, it means that the model suffers and that it is not able to select variables.

Back to article page