Fig. 5

Results of fitting the Unsplit and Split LK-DFBA models to the E. coli data. Solid lines represent the median penalized relative sum-of-square error (prSSE) for each modeling approach and dotted lines represent the median absolute deviation. a CoV = 0.10. b CoV = 0.20. nT is the number of time points used to fit each model. At higher noise and higher data frequency, the split method performs better even when penalized for having additional fitted parameters