Skip to content

Releases: gtathelegend/BehaviourSim

BehaviorSim v1.0.0

Choose a tag to compare

@gtathelegend gtathelegend released this 12 Sep 11:08

BehaviorSim v1.0.0

BehaviorSim v1.0.0 is the first stable release of a general-purpose Python library for generating synthetic sequential behavioral data with configurable hidden states, observable behavior, ground-truth labels, calibration, causal feature engineering, and reproducible simulation.

Highlights

  • Configurable behavioral simulation

    • Arbitrary behavioral states and transition matrices
    • Per-state feature distributions
    • History-dependent transition rules
    • Deterministic and reproducible sequence generation
  • Multi-profile simulation

    • Independent behavioral profiles
    • Profile-specific emissions and transition dynamics
    • Isolated sequence seeds
  • Domain presets

    • Education
    • Mobile applications
    • Healthcare
    • Finance
  • Empirical calibration

    • State-proxy extraction
    • Transition-matrix fitting
    • Emission parameter estimation
    • Profile fitting and clustering
    • Calibration validation
  • Causal behavioral feature engineering

    • Rolling statistics
    • Streaks
    • Sessionization
    • Transition dynamics
    • Strictly causal feature construction
  • Visualization

    • State occupancy
    • Behavioral trajectories
    • Distribution comparisons
    • Calibration diagnostics
    • PNG, PDF, and SVG figure export
  • Configuration & CLI

    • YAML/JSON configuration
    • Configuration validation
    • behaviorsim command-line interface
    • Python API

Quality & Validation

The v1.0.0 release was validated through:

  • 634 tests passed
  • 5 tests skipped
  • 4 pre-existing legacy warnings
  • Python 3.9–3.12 compatibility coverage
  • Linux and Windows CI coverage
  • Wheel and source-distribution builds
  • Isolated wheel installation and execution
  • CLI smoke testing
  • Reproducibility verification
  • Package metadata verification

The legacy CLSI-Adapt evaluation code remains preserved independently from the BehaviorSim release.

Performance

Current in-memory generation throughput is approximately 6,000 interactions/second under the validated benchmark environment.

The original PRD target of 1 million interactions in under 10 seconds remains an aspirational future optimization target and is not treated as a v1.0.0 correctness or release blocker.

Installation

BehaviorSim is packaged as a standard Python distribution with Python 3.9+ support.

The release artifacts include:

  • behaviorsim-1.0.0-py3-none-any.whl
  • behaviorsim-1.0.0.tar.gz

Release Status

Stable — v1.0.0

This release represents the completed first production milestone of BehaviorSim.