Added ranking prior to permutation testing

[sarthakpati]

Fixes #9

Also adds a better default for CLI

[Copilot]

Pull Request Overview

This PR fixes issue #9 by adding ranking functionality prior to permutation testing and improves the CLI default value. The changes ensure that data is properly sorted by cumulative rank before performing permutation tests, and provides a better default for the metrics-for-reversal CLI parameter.

• Sorts ranks_per_metric by cumulative_rank before sanitization in permutation testing
• Resets the dataframe index after sorting to maintain proper ordering
• Changes CLI default for metrics-for-reversal from None to empty string

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
pyranker/ranker.py	Adds sorting by cumulative_rank and index reset before permutation test sanitization
pyranker/cli/run.py	Changes default value for metrics-for-reversal parameter from None to empty string

Comments suppressed due to low confidence (1)

pyranker/ranker.py:129

• [nitpick] The variable name 'ranks_per_metric_sorted' is verbose and inconsistent with the existing naming pattern. Consider renaming to 'sorted_ranks' or 'ranks_sorted' to be more concise and align with the existing 'ranks_per_metric_sanitized' naming pattern.

        ranks_per_metric_sorted = self.ranks_per_metric.sort_values(by="cumulative_rank")

[sarthakpati]

Here are the results on the sample data:

• pvals without sorting (2772d80):

method,m1,m2,m3,m4
m1,0.0,0.913,0.906,0.548
m2,0.0,0.0,0.434,0.112
m3,0.0,0.0,0.0,0.112
m4,0.0,0.0,0.0,0.0

• pvals with sorting (0bbbde4):

method,m1,m2,m3,m4
m1,0.0,0.429,0.091,0.08
m2,0.0,0.0,0.101,0.057
m3,0.0,0.0,0.0,0.384
m4,0.0,0.0,0.0,0.0