Closed hmgaudecker closed 4 years ago
@janosg: This happens for the x, p - parameters. In the broader picture, would it not be more useful to use only name1 (x) and both name1 / name2 (p) ?
I remember that issue too. Probably the easiest solution is to add a load_params
method. I also agree with that we should use name1 and name2 for p.
I think this should be fixed at the point of exporting the parameters -- or is there a specific reason why it should be 0
for the name as opposed to the empty string?
The first 0 is for the period. The second one enumerates the mixture elements. The period is redundant because x is only estimated in period 0. I guess this is a case of practicality beats purity
and we should omit the first zero.
Ah, okay. I would vote for purity and being really explicit then.
mixture_0
, mixture_1
, ...
A related problem is that empty strings get replaced by np.nan when read from csv. I will replace all empty strings in the index by "-".
e.g., I have some stuff where start_params_helpers() produces
among other things. Note the
0
integer forname1
. The problem is that if you save this as csv, change the start values, and read it back into Pandas,read_csv
will convert the integer into a string andfit()
will complain with:Not the worst thing in the world, but a bit annoying...