Skip to main content
Fig. 6 | Microbiome

Fig. 6

From: Proportion-based normalizations outperform compositional data transformations in machine learning applications

Fig. 6

Pairwise p-values indicate that lognorm performs significantly better than every other transformation. The y-axis shows log10 of the p-value, and the x-axis shows how well each transformation performed against the other 11 transformations. A positive value indicates that the average accuracy of the given transformation is higher than others, and a negative p-value indicates that the average accuracy is lower than the others. The area above the solid green line represents the sample space where the transformation is significantly better, and the area below the solid red line represents the area where points are significantly worse

Back to article page