requested edits added

[aryakrekhi]

No description provided.

[codecov]

Codecov Report

Merging #161 (2e98627) into main (348aec9) will decrease coverage by 0.52%.
The diff coverage is 62.85%.

@@            Coverage Diff             @@
##             main     #161      +/-   ##
==========================================
- Coverage   51.28%   50.76%   -0.53%     
==========================================
  Files          16       16              
  Lines         969      979      +10     
==========================================
  Hits          497      497              
- Misses        472      482      +10     

Flag	Coverage Δ
unittests	50.76% <62.85%> (-0.53%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
de/linearModel.py	26.66% <14.28%> (-6.67%)	⬇️
de/impute.py	49.01% <40.00%> (-11.92%)	⬇️
de/fancyimpute/soft_impute.py	86.95% <75.00%> (-9.97%)	⬇️
de/factorization.py	100.00% <100.00%> (ø)
de/fancyimpute/solver.py	61.85% <0.00%> (+3.09%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 348aec9...2e98627. Read the comment docs.

[aarmey]

@aryakrekhi @Farnazmdi ok, this now has moved imputation to within the fitting loop. I think this will be much faster and have similar results to before. Please try it out and see if I'm right.

[Farnazmdi]

Thank you! It certainly is faster, however, I ran the figure for comparing linear and nonlinear fitting, they don't seem to be significantly different...

[aarmey]

Is this a within-row correlation like we talked about? I forget if that's already been merged.

[Farnazmdi]

Yes, it is reflected in line 12 of impute.py file.

[aarmey]

How are you running the linear impute?

[Farnazmdi]

Ah.. that's the reason.. I should have changed it based on the new implementation. I think it doesn't actually run the linear at all, both boxplots are from nonlinear..
I will work on this and get back to you.

[Farnazmdi]

I edited the linear model in the new imputation setting and ran the correlation figure again. The changes are pushed, the figure took 144473 seconds (40 hours) to run -- slow because of linear model!! And the linear model looks better in most of them!