Fig. 8
From: A machine learning approach to predict metabolic pathway dynamics from time-series multiomics data
Predictions improve with more training data sets. The machine learning algorithm was used to predict kinetic models for varying sizes of training sets (2, 10, and 100 virtual strains in blue, red and black). Ten unique training sets were used for each size to show prediction variability (transparency) for each training set size. All models converge towards the actual dynamics with the 100 strain models in closest agreement. Standard deviations (shown by the transparency) also decrease markedly as the size of the training set increases
