Add two indicators to determine the ranks

[sshojiro]

This PR (derived from 0.3.X) includes following features:

1. IND (INDicator) and ROD (ratio of derivatives) are added
2. Example notebook for the indicators is added
3. IND is computed with only numpy svd module, aiming at a reduction of dependency

[codecov[bot]]

Codecov Report

Merging #7 into 0.3.X will decrease coverage by 0.8%. The diff coverage is 87.12%.

@@            Coverage Diff             @@
##            0.3.X       #7      +/-   ##
==========================================
- Coverage   98.39%   97.58%   -0.81%     
==========================================
  Files           9       13       +4     
  Lines        1306     1407     +101     
==========================================
+ Hits         1285     1373      +88     
- Misses         21       34      +13

Impacted Files	Coverage Δ
pymcr/metrics.py	100% <ø> (ø)	:arrow_up:
pymcr/tests/test_condition.py	100% <100%> (ø)
pymcr/tests/test_rank.py	100% <100%> (ø)
pymcr/condition.py	100% <100%> (ø)
pymcr/rank.py	59.37% <59.37%> (ø)

---

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 dffc120...e043896. Read the comment docs.

[CCampJr]

Updates:

• Moved _scale in rank.py to standardize in condition.py as this is pretty common and useful conditioning step
• Added a stand-alone PCA function in rank.py
  • Note: PCA is not standardized but just mean-centered
  • This PCA can do full PCA or truncated
  • Note: scikit-learn does a stochastic version of PCA which should agree to all but the last component
• Added tests for standardize and pca

Still needing to be done:

• Tests for ind and rod
• Implement the new pca into the ind function
• Update docstrings for ind and rod
• Update the Jupyter notebooks to comply with any changes to the ind and rod functions

[CCampJr]

Hi @sshojiro ,

Just checking if you've seen the alterations I made. It should give the same results as what you first prepared.

Thanks!

[sshojiro]

@CCampJr Well, I've found a mistake in IND definition. I'm working on an implementation of IND function and test for both ind and rod now.

I've modified the docstring into ind and rod. Sorry for this late response.