Fixes #247 - LSA should not sqrt singular values

[singularperturbation]

• Take out sqrt of singular values in LSA$fit_transform
• Add tests for LSA
  • LSA$fit_transform should match manual decomp
  • Explained variance when using max dimension size for decomp (max number of rows and columns - 1) should be close to 1.0

[singularperturbation]

Second commit was to update LSA$transform to work in the same way as LSA$fit_transform

[singularperturbation]

Thanks for the quick response! I'll review and make these changes this weekend.

[codecov[bot]]

Codecov Report

Merging #248 into master will decrease coverage by 0.11%. The diff coverage is 71.42%.

@@            Coverage Diff            @@
##           master    #248      +/-   ##
=========================================
- Coverage   80.02%   79.9%   -0.12%     
=========================================
  Files          38      38              
  Lines        2343    2359      +16     
=========================================
+ Hits         1875    1885      +10     
- Misses        468     474       +6

Impacted Files	Coverage Δ
R/model_LSA.R	78.72% <71.42%> (-8.38%)	:arrow_down:

---

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 2626bcd...5f5bff3. Read the comment docs.

[dselivanov]

Perfect, thanks!