Need to add dependence to iterable dataset. How to do this is still in question since i would like to keep the sampler as dynamic. Best bet is to still load in Truth data but on-the-fly load in patches of forecast data. Mapping will also need to be incorporated for n_workers.