It would be useful to have additional result files handy for the calibrate operation.
We can extrapolate on the current results and store these additional things into S3 along with the other "result_files".
There additions are:
[ ] A sampling of the dill file that produces a distribution histogram that we can use to get a sense of what were the calibrated parameter values.
This is meant to generate the histograms
param a
param b
param ...
sample value
sample value
...
sample value
sample value
...
...
...
...
[ ] Errors outlining how far each sample as away from the ground truth, if applicable
This is meant to generate the error distribution and raincloud-plot
sample
state a
state b
state ...
1
error
error
...
2
error
error
...
...
...
...
...
Note 1: Probably talk to @liunelson about point-1.
Note 2: The formats may be subject to change
It would be useful to have additional result files handy for the calibrate operation. We can extrapolate on the current results and store these additional things into S3 along with the other "result_files".
There additions are:
This is meant to generate the histograms
This is meant to generate the error distribution and raincloud-plot