feat: Implement control subtraction for LASSO behavioral prediction and add corresponding tests

[colehanan1]

check if lasso working right it seems to be!

Summary by CodeRabbit

• New Features

  • Added diagnostic tool for LASSO-based behavioral prediction analysis with support for multiple prediction modes and control condition handling.
  • Generates comprehensive reports including per-condition metrics, coefficient summaries, collapse flags, and reproducibility checks in CSV, JSON, and markdown formats.

• Tests

  • Added regression tests to validate behavioral prediction component integrity and prediction consistency across repeated fits.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Caution

Review failed

The pull request is closed.

📝 Walkthrough

Walkthrough

This pull request introduces a comprehensive diagnostic framework for LASSO-based behavioral predictions. It adds a new Python script that orchestrates multi-condition and multi-mode diagnostic runs, aggregates results into various output formats (CSV, JSON, Markdown), and includes regression tests to validate newly exported utility functions for receptor ablation and fixed-scaler fitting.

Changes

Cohort / File(s)	Summary
Configuration .gitignore	Added diagnostics/ directory to ignore list (1 line addition)
Diagnostic Script scripts/diagnose_lasso_regressions.py	New 428-line script introducing: RunResult dataclass to encapsulate per-run outcomes; run_mode() function executing LASSO behavioral predictor with configurable parameters; argument parsing with custom validators; statistical computation helpers; main orchestration logic aggregating results across conditions/modes; output generation for CSV/JSON/Markdown formats with metrics, top coefficients, reproducibility checks
Regression Tests tests/test_lasso_behavioral_prediction.py	Added 91 lines of new tests: TestLassoBehavioralPredictorRegressionChecks class with two test methods validating that receptor ablation doesn't mutate baseline matrices and that delta predictions remain non-constant under varied configurations

Sequence Diagram(s)

sequenceDiagram
    actor User
    participant CLI as Command Line
    participant Parser as Argument Parser
    participant Orchestrator as Main Orchestrator
    participant Predictor as LassoBehavioralPredictor
    participant Stats as Statistics Engine
    participant FileIO as File I/O

    User->>CLI: Execute script with arguments
    CLI->>Parser: Parse conditions, modes, lambda_range
    Parser-->>CLI: Validated arguments
    CLI->>Orchestrator: Initiate diagnostic run
    loop For each condition and mode
        Orchestrator->>Predictor: run_mode (condition, mode, params)
        Predictor->>Predictor: fit_behavior
        Predictor->>Stats: Compute y_stats, x_stats, pred_stats
        Stats-->>Predictor: Statistics (mean, std, collapse flags)
        Predictor-->>Orchestrator: RunResult + details dict
    end
    Orchestrator->>Stats: Aggregate metrics across runs
    Orchestrator->>FileIO: Write per_condition_metrics.csv
    Orchestrator->>FileIO: Write per_condition_top_coefs.csv
    Orchestrator->>FileIO: Write collapse_flags.csv
    Orchestrator->>FileIO: Write per_condition_yx_stats.json
    Orchestrator->>FileIO: Write reproducibility_check.csv
    Orchestrator->>FileIO: Write SUMMARY.md
    FileIO-->>User: Output files generated

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Possibly related PRs

• Feature/lasso-ablation-analysis #2: Introduces apply_receptor_ablation and fit_lasso_with_fixed_scaler functions that are directly utilized in the new regression test class for validating mutation prevention and fixed-scaler behavior.

Poem

🐰 Hoppity hop, diagnostics flow,
LASSO runs in rows, statistics glow,
Ablation's ablaze, predictions align,
CSV dreams and JSON shine! ✨

✨ Finishing touches

• 📝 Generate docstrings

---

📜 Recent review details

Configuration used: defaults

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between a3dfa99 and 36c6149.

📒 Files selected for processing (3)

• .gitignore
• scripts/diagnose_lasso_regressions.py
• tests/test_lasso_behavioral_prediction.py

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}