Deal with missing data?

[SandorAlbert]

Is the Optimizer class capable of dealing with missing data? For example when one factor/variable to be optimized has a np.nan entry for whatever reason. Or when you would like to extend the model with new data but without some variables since they are not of interest anymore. Thanks in advance!

[RuneChristensen-NN]

1: The optimizer cannot handle NaN entries in the training/input data. You have to do the data cleaning, drop factors, or whatever else you might like to do as part of the pre-processing of data. Normalization of data is handled by the optimizer such that you can keep input and output in real space.

2: If you have a dataset and wish to drop factors, the best way is to drop the factors in your dataset, initialize a new Optimizer object, define the space in the new optimizer such that it fits your current wanted factors, and train the new optimizer using tell([input data])

Hope that helps