As an aside, some machine learning techniques use a fourth set of data, called a validation set, which is used during training to avoid overfitting.
The gradient model of bacterium can accurately predicted the number of bacteria. The mean RSD for the validation set were 6.61% and 4.59%.