I am using an
LSTM to predict a time series chart from multiple other time series charts as input features. The problem is that some of these input charts have much fewer data. For example for some input charts, data starts from 1990s while for other input charts, data starts from 2009.
As can be understood here, if for missing values, a placeholder such as
-1 is inserted, a neural network is able to learn that those values should be ignored.
Will this hold true for my situation where I would have a large amount of
-1 values where data starts from 2009?