fix x range when evaluating curves in sample function #11
Conversation
windisch
left a comment
windisch
left a comment
There was a problem hiding this comment.
Good catch - looks good to me. Just a bit worried about complexity due to code duplicates. Added a suggestion - freel free to explore this further!
driftbench/data_generation/sample.py
Outdated
| # Apply some random noise on the base values | ||
| x0 = base_latent_information.x0 + rng.normal(size=len(base_latent_information.x0), scale=x_scale) | ||
| y0 = base_latent_information.y0 + rng.normal(size=len(base_latent_information.y0), scale=y_scale) | ||
| x0 = base_latent_information.x0 + rng.normal( |
There was a problem hiding this comment.
Any idea to remove the code duplicates? We already had them, but now it gets more nasty...
One idea may be to have a latent dict which keys y0, x1, like:
latent["x0"] = getattr(base_latent_information, 'x0') + rng.normal(
size=len(getattr(base_latent_information, 'x0')), scale=scale["x"]
)What do you think?
|
@windisch Followed your suggestion, looks cleaner now. May also think about changing |
| y2 = base_latent_information.y2 + rng.normal(size=len(base_latent_information.y2), scale=y_scale) | ||
| latent_information.append(LatentInformation(y0, x0, y1, x1, y2, x2)) | ||
| latent_dict = {} | ||
| for xi in xis: |
There was a problem hiding this comment.
this may could also be a single for loop :-)
for v in xis+yis:with x_scale and y_scale somehow accessed via v[0] and
scale = {'x': x_scale, 'y': y_scale}But fine for me for now!
Done in this PR
x_iinsample_curveswere considered for the evolution x range.