Open chriselrod opened 1 year ago
Merging #191 (f0d7d4d) into main (94e9907) will decrease coverage by
6.96%
. The diff coverage is21.90%
.
@@ Coverage Diff @@
## main #191 +/- ##
==========================================
- Coverage 73.36% 66.40% -6.96%
==========================================
Files 7 7
Lines 428 512 +84
==========================================
+ Hits 314 340 +26
- Misses 114 172 +58
Impacted Files | Coverage Δ | |
---|---|---|
src/EasyModelAnalysis.jl | 100.00% <ø> (ø) |
|
src/datafit.jl | 48.07% <21.90%> (-11.61%) |
:arrow_down: |
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
I am running fits, hoping it converges. Haven't looked at results yet. In the following example's current form, it is too slow for CI:
Weights are currently constrained to sum to 1, but they are also allowed to be negative, e.g. if you have two models, you could end up with a solution like
2 .* sol1 .- sol2
. We could place it on a simplex instead, but I figured to start with, it'd be interesting to see if it does like such solutions where you subtract a result.My concern is that this could be too much freedom/allow for creative overfitting. I won'd want to see solutions like