DeciShift v0.3.1 — Trust and evaluation hardening

DeciShift v0.3.1 — Trust and evaluation hardening

This release strengthens the trust layer around decision-change evidence while adding reproducible public-data evaluation.

Highlights

• Public trained DecisionFlow examples using scikit-learn Wine and XGBoost Digits datasets.
• Machine-generated decision-change case study: improved model metrics while 4.17% of individual actions changed, with exact component attribution and a governed cohort contract returning BLOCK.
• Commit-tied 10K/100K exact-vs-sampled attribution benchmark snapshots with cache reuse, wall time, Python allocation and convergence diagnostics.
• Legacy DecisionPipeline caches bound to input-content fingerprints with defensive-copy and caller-input isolation protections.
• Hybrid replay invalidation when record content changes.
• Enforced branch-aware coverage floor raised to 80% with targeted trust-path regression tests.
• GitHub Actions refreshed and pinned to immutable commit SHAs; release publication continues through PyPI Trusted Publishing/OIDC.
• Expanded security and reproducibility documentation, including integrity-vs-authenticity limits.

Release-source note

The v0.3.1 tag is created from exact commit 082250abc1c0a82a41480949594168bb56f323ad, which passed the repository's full post-merge test matrix. The repository owner explicitly authorized this release without main branch protection; issue #7 tracks enabling branch-level enforcement for future releases.

Scientific scope

DeciShift attribution is software-counterfactual attribution. It does not establish real-world causality, safety, fairness, compliance or production fitness.

DeciShift v0.3.0 — Composable Decision Flows

DeciShift v0.3.0 adds Composable Decision Flows while preserving the existing DecisionPipeline API and v0.1/v0.2 behavior.

Added

• DecisionNode, DecisionFlow, and FlowTrace for deterministic row-aligned DAG decision systems.
• Branching, merging, multiple models/policy stages, and categorical/multi-action final decisions.
• Deterministic DAG validation and canonical SHA-256 topology fingerprints without NetworkX.
• Graph-aware cache reuse and descendant invalidation metrics.
• Structural impact analysis separated from observed action changes and causal claims.
• Multi-action transition summaries, matrices, action distributions, and Decision Contracts.
• Numeric flow-attribution targets: candidate_action_support and change_from_baseline.
• Exact and adaptive approximate node/group attribution with sampling uncertainty.
• Flow interaction analysis including interaction-only categorical transitions.
• mode: flow YAML support and decishift graph text/Mermaid inspection.
• DecisionFlow evidence schema 2.0 with backward-compatible evidence loading/verification.
• Synthetic equipment-maintenance triage example and CPU benchmark harness.

Attribution correction

• Separates Shapley efficiency from Monte Carlo precision/convergence semantics.
• Adds adaptive permutation sampling with minimum/maximum permutations, batches, CI-width targets, confidence level, and deterministic seeds.
• Preserves fixed-permutation mode and existing public APIs.

Compatibility and scientific scope

• Existing v0.1/v0.2 public APIs, binary reports, contracts, evidence, and CLI behavior remain supported.
• Categorical actions are never numerically subtracted or silently ordinal-encoded.
• Structural reachability is not causal impact.
• Software-counterfactual attribution does not establish real-world causality.
• Topology-changing node/group hybrid attribution remains unsupported rather than fabricated.

Install

pip install --upgrade decishift==0.3.0
decishift demo

DeciShift v0.2.0 — Trustworthy Decision-Change Evidence

DeciShift v0.2.0 adds a Trustworthy Decision-Change Evidence layer while preserving the v0.1 public API and CPU-first, local-first design.

Added

• Streaming Monte Carlo uncertainty for approximate component attribution, including standard errors and configurable confidence intervals.
• Attribution diagnostics with Shapley efficiency residual checks and warnings.
• Stable component provenance and strict reproducibility mode.
• Evidence schema 1.0, deterministic fingerprints, manifests, SHA-256 integrity verification, and decishift verify.
• User-defined Decision Contracts with deterministic decishift gate exit codes for CI.
• Stronger record identity, finite-output, row-order, and callable-invocation validation.
• Dynamic YAML threshold factories/objects while retaining numeric thresholds.
• Safer cohort discovery with Wilson intervals, coverage, global context, and excess flip rate.
• Optional binary outcome correctness-transition analysis.
• Decision Fragility diagnostics for threshold proximity and boundary crossing.
• Fully local self-contained HTML reports.
• Saved-run comparison with decishift compare-runs.
• Expanded benchmarks, regression coverage, Ruff, build, and Twine checks.

Compatibility

Existing v0.1 public imports remain supported, including DecisionPipeline, compare_pipelines, compare_predictions, exact_attribution, approximate_attribution, and pairwise_interactions.

Install

pip install --upgrade decishift==0.2.0
decishift demo

Scientific scope

DeciShift attributes observed software-output changes across a versioned executable decision pipeline. Component attribution is software-counterfactual attribution and does not establish external real-world causal effects. A Decision Contract PASS means only that the user-configured checks passed; it is not proof of safety, fairness, compliance, or correctness.

DeciShift v0.1.0

DeciShift v0.1.0 is the initial public release.

DeciShift is a CPU-first, local-first framework for explaining why individual decisions changed between versions of a machine-learning decision pipeline.

Added

• CPU-first, local-first comparison of baseline and candidate ML decision pipelines.
• Version-aware pipeline components for features, model, calibration, threshold/policy, and deterministic rules.
• Record-level decision diffs, score deltas, threshold margins, and flip direction.
• Exact Shapley counterfactual component attribution for small changed-component sets.
• Deterministic permutation approximation for larger attribution problems.
• Baseline-anchored pairwise interaction analysis, including interaction-only decision flips.
• Predictions-only analysis with explicit insufficient-evidence handling for component attribution.
• Numeric and categorical cohort analysis with minimum cohort-size controls.
• Saved local runs and individual record explanations.
• Terminal, JSON, and Markdown reporting.
• Framework-agnostic model interface with optional scikit-learn, XGBoost, and LightGBM support.
• Synthetic equipment-maintenance demonstration.
• Reproducible CPU benchmark harness.
• Regression tests and GitHub Actions CI.
• Apache-2.0 licensing.
• CITATION.cff metadata for software citation and Zenodo archival.

Install

After publication to PyPI:

pip install decishift

Then run:

decishift demo

Scientific scope

DeciShift reports observed decision differences and counterfactual attribution across executable software components. These results should not automatically be interpreted as real-world causal effects.