Python analysis module: z-score severity thresholds are hardcoded

[AnupamKumar-1]

Called out in the README's "Anomaly detection: Node vs Python" table — the TS core supports a configurable z-score threshold, Python doesn't. In python/obyflow-python/obyflow/analysis/stats.py:

def classify_severity(z_score: float) -> DeviationSeverity:
    abs_z = abs(z_score)
    if abs_z < 1:
        return "none"
    if abs_z < 2:
        return "low"
    if abs_z < 3:
        return "medium"
    return "high"

The 1/2/3 cutoffs are fixed. Should let callers pass their own.

• Add optional threshold params to classify_severity, e.g. low_threshold=1.0, medium_threshold=2.0, high_threshold=3.0
• Thread them through as kwargs in detect_ml_anomalies (analysis/anomaly.py), the only caller
• Add tests in test_analysis_stats.py / test_analysis_anomaly.py for non-default thresholds
• Update the README table

[Ochibueze474]

Hi, I'd like to work on this one. I've worked with default arguments in Python before, so this pattern makes sense to me.
I'll add the threshold parameters to classify_severity, thread them through detect_ml_anomalies, add tests for non-default thresholds, and update the README table as described.

Will share progress here before opening a PR.

[AnupamKumar-1]

Hi, I'd like to work on this one. I've worked with default arguments in Python before, so this pattern makes sense to me. I'll add the threshold parameters to classify_severity, thread them through detect_ml_anomalies, add tests for non-default thresholds, and update the README table as described.

Will share progress here before opening a PR.

Sounds good, I’ve assigned it to you. Let me know if you need help.

[Ochibueze474]

Just opened the PR for this one, hope the approach makes sense!
Thanks for assigning it to me and for the earlier reply, made it easy to get started.