The ability to reconstruct missing values measures the quality of a model: its ability to generalise, memorise and represent. The nonlinear factor analysis model turned out to perform well in generalisation and representation while its ability to memorise is limited due to the small number of parameters in the model. NFA performed better than FA in all the experiments, but with large number of factors, the current NFA algorithm becomes computationally expensive. We conclude that nonlinear factor analysis (NFA) is best suited for fairly strongly nonlinear problems with an intrinsic dimension of the order of ten.