HTML good first issues

The `faircode` tool crashes with a `ZeroDivisionError` when the `--n-permutations` argument is set to 0 during benchmarking. This occurs because the `permutation_test` function divides by `n_permutations` without checking if it's zero, and the error handling in the benchmark command doesn't catch this specific exception. The suggested fix involves validating the input at the argument parsing level or adding `ZeroDivisionError` to the exception handling.

This issue outlines a large-scale effort to curate and categorize 3,079 rare diseases from a prioritized list. The categorization is based on matching MONDO IDs and various metadata fields, with specific examples provided for neurodevelopmental, neurodegenerative, neuroimmune, and cardiac disease groups. The goal is to enhance the existing knowledge base by organizing these diseases into relevant categories.

The Makefile's comment for the 'audits' target incorrectly states that 'run-audits' is a required check. This is not true according to branch protection rules and other documentation, and the fix should align with the resolution of issue #605 regarding the actual requirement status of 'run-audits'.

The README.md file incorrectly states there are six projects when there are actually seven. Additionally, the list of projects for swapping out the placeholder is missing 'Tenant Screening'. The fix involves updating the count to seven and adding the missing project to the list.

The documentation for the Simpsons Paradox explainer has a minor arithmetic error in the stratified-rate table. The 'Separated' row incorrectly states the gap as +9.8 pp when the actual calculated value, based on the provided Python script and dataset, is approximately +9.7 pp. The fix involves correcting this displayed percentage point difference.

The `index.html` file incorrectly displays a 97% bias reduction figure in its hero section, which actually belongs to a different audit (AI Fair Recruitment). The first two stats in the hero section refer to COMPAS, but the third stat (97%) misrepresents COMPAS's actual bias reduction of 82%. The fix involves either correcting the third stat to reflect COMPAS's true reduction or clarifying that it's a cross-project statistic.

The `profiler.html` file incorrectly states that there are 'six bias audits' when there are actually seven. This is a documentation error that needs to be corrected by changing the text from 'six' to 'seven'. The issue is similar to previously fixed documentation bugs.

The documentation for the ROC curve and AUC in the `roc-curve-auc.md` file incorrectly states that the COMPAS example model predicts actual recidivism. In reality, the model predicts COMPAS's own risk score, not real reoffense, as indicated by the `audit.yaml` configuration. The fix involves rewording the description to accurately reflect what the model predicts without altering the existing AUC values.

The documentation for predictive parity in the repository incorrectly references a 'decile_score' column and a 'two-year re-offense' target, neither of which exist in the provided COMPAS CSV file. The actual column is 'DecileScore' and the target is 'ScoreText' as defined in the audit configuration. This issue requires updating the documentation and code to reflect the correct schema and target.

The documentation for the base-rate fallacy explainer incorrectly references a 'race' and 'two_year_recid' schema that does not exist in the provided COMPAS CSV file. The actual column names are 'Ethnic_Code_Text' and the recidivism outcome needs to be derived from 'ScoreText' and 'DisplayText' columns, similar to previous fixes for other explainers. The fix involves updating the code example to use the correct schema and regenerating the output.

The documentation for bootstrap confidence intervals incorrectly states that `faircode/significance.py` is a frozen file. This is no longer true, as the repository's freeze has been lifted and the file itself does not claim to be frozen. The fix involves updating the documentation to reflect the current state or potentially importing the `significance.py` file directly.

The `compare_to_terminal()` and `compare_to_html()` functions in `faircode/report.py` incorrectly display a 'no drift' result for dimensions where the comparison was skipped due to a kind mismatch (e.g., string vs. numeric age). This creates a self-contradictory report, showing zero drift metrics alongside a flag indicating the comparison was skipped. The fix involves updating these renderers to display a dedicated 'comparison skipped' badge, similar to what's already implemented in the JavaScript assets.

The `faircode benchmark` command crashes with an `IndexError` when `--n-resamples` is set to 0. This is because the code does not validate that `--n-resamples` must be a positive integer, allowing 0 to be passed to a NumPy function that subsequently fails. The fix involves adding validation for this argument.

The `faircode` benchmark command leaks raw `FileNotFoundError` tracebacks when a specified manifest file does not exist. This occurs because the error handling in `run_benchmark` does not include `FileNotFoundError`, contrary to the intended behavior of providing clean error messages. The fix involves adding `FileNotFoundError` to the exception handling to ensure consistent error reporting.

The `profile` command in `faircode.cli.py` lacks a guard against reading both the primary dataset and a reference/proxy hints dataset from standard input simultaneously. This leads to a misleading error message when stdin is consumed twice, as the second read returns nothing. The fix involves adding a check similar to the `compare` command to prevent this scenario and provide a clearer error.

The `_labelize` function in `faircode/proxy.py` silently drops rows with non-numeric age sentinels (like 'unknown') during the chi-squared proxy test. This leads to an undercounted sample size and potentially biased results, as the `_intersections` function in `faircode/profiler.py` already fixed this issue for a similar scenario.

The branch protection rules for the 'main' branch on GitHub do not enforce required CI checks, despite documentation in CONTRIBUTING.md and workflow comments stating otherwise. This discrepancy allows PRs with failing CI to be merged, and a workflow file is being kept as a no-op under the false assumption that it's a required check.

This issue proposes adding a new 'Stat Counter Card' component to the EaseMotion CSS gallery. The component will feature a metric card with an animated number and a mini bar chart that animates on hover. The implementation involves creating a demo HTML file and a CSS file for the component's styling.

This issue proposes adding a new feature to the EaseMotion-css repository, specifically focusing on advanced component iteration using pure CSS. It's labeled as a "help wanted" and "good first issue," suggesting it's a good starting point for contributors and aims to enhance the repository's capabilities.

New users are being directed towards `jpm` as the primary package manager due to unclear documentation, overlooking other potentially valuable options. The issue proposes updating the documentation to highlight these alternatives and guide users towards making informed choices.

This issue tracks the migration of several membership-related Odoo modules to version 19.0. It provides a checklist of modules that need to be updated and links to relevant documentation for the migration process.

This issue proposes adding a new animated accordion component to the repository. The component should feature smooth transitions for expanding/collapsing content and a rotating chevron icon. This is intended to enhance the user experience for displaying structured content or FAQs.

This issue proposes adding a new advanced component, implemented purely with CSS, to the EaseMotion-css repository. The goal is to introduce comprehensive functionality, likely for version 17 of the library.

This feature request proposes adding a new SCSS animation mixin pattern called 'Gradient Flow Shift Color' to the EaseMotion core animation engine. The goal is to create a hardware-accelerated, 60 FPS animation that shifts colors within a gradient, with configurable timing and support for reduced motion preferences.

This issue proposes adding a new example to the `submissions/examples/` directory: an animated banking dashboard. This example would showcase advanced CSS techniques like 3D card effects, financial data visualizations, transaction lists, and savings goals, providing a comprehensive fintech interface demonstration.

The web profiler's JavaScript engine silently drops data categories named '__proto__' because it uses a plain object as a hashmap, leading to incorrect group counts and statistics. This issue corrupts data integrity in the web profiler's dimension analysis. The proposed fix involves using a `Map` or `Object.create(null)` for hashmap implementations in `profiler-engine.js`.

This issue proposes to refactor the product deletion process. Instead of deleting individual images associated with a product, all product images should be moved to a 'deleted' folder when the product itself is marked as deleted. This change aims to preserve image data temporarily while ensuring the product is no longer visible.

This issue highlights a discrepancy in the `CONTRIBUTORS.md` file where the Git identity map for `@TanishGoyal-Dev` doesn't explain why they are missing from the GitHub contributors graph. The problem is identical to that of `@Ayaan-20-11`, whose entry correctly explains their absence due to an unlinked commit email. The proposed fix is to add a similar explanation to `@TanishGoyal-Dev`'s entry.

The documentation for the 'Benefits Denial' audit incorrectly omits the 'Age' attribute from its results tables, despite the underlying scripts calculating and reporting a gap for it. The fix involves adding the missing 'Age' row to both the root README and the audit's specific README, using the computed figures.

The `disparate_impact_ratio` metric in `faircode/metrics.py` incorrectly hardcodes the significance threshold to p < 0.05. This ignores the `confidence` parameter, which is used by other fairness metrics to dynamically determine significance. The fix involves updating the `_ratio_report` function to use the confidence parameter, ensuring consistent significance flagging across all metrics.