pySEQTarget 0.13.1

• Add Tom to maintainers

pySEQTarget 0.13.0

• Preserve eligible_col during expansion when referenced by weight_eligible_colnames
• Validate weight_eligible_colnames entries exist in data upfront
• Skip bootstrap replicates that fail with singular Hessian
• Silence non-interactive backend plt.show warnings in test suite
• Bump astral-sh/setup-uv to v8.1.0

pySEQTarget 0.12.9

• Fix expansion bug where subjects experiencing outcome early were carried forward with post-outcome rows

pySEQTarget 0.12.8

Hopefully remove <factory> from rendered docs and some optimizations and speed ups.

pySEQTarget 0.12.7

• In the bootstrapping use paired per-iteration estimates and monitor both risk difference and log risk ratio - there's no need for the delta-method.
• Document list field defaults in SEQopts docstrings to fix <factory> display on readthedocs site https://pyseqtarget.readthedocs.io/en/latest/sequential/seqopts.html
• Fixes the incorrect display of ncores=1 on the seqopts readthedocs site
• Fixes the display of summary and retrieve_data on https://pyseqtarget.readthedocs.io/en/latest/sequential/seqoutput.html
• Fixes the default of weight_preexpansion in the advanced models vignette

pySEQTarget 0.12.6

• Fixes SEQoutput.plot() to display figure by calling plt.show() instead of printing repr
• Fixes SEQuential.plot() to display figure by calling plt.show()
• Re-enables Python 3.10 as I think the lifelines package works under this once more
• Adds some more of the validation checks in the R package
• Amends the default values of weight_preexpansion, plot_type, and ncores to match those in SEQTaRget

pySEQTarget 0.12.5

• Fix Nan proliferation in data
• Fixed casting in predictions
• Changed probability warning to valueError

pySEQTarget 0.12.4

• Guard against None models in SEQoutput.summary(). When a treatment-level subset lacks sufficient unique values, _fit_numerator/_fit_denominator append None to the fits list. summary() now skips None entries to avoid AttributeError.

pySEQTarget 0.12.3

This,

• removes the unneeded import in the README example (somehow missed that before)
• amends some typos in arguments and docstrings
• adds a check for perfect separation
• handles no variation weight models
• bumps some actions in GHA workflows
• allow readthedocs to recover version information from the toml tag

pySEQTarget 0.12.2

• Make main estimate and bootstrapping deterministic for a seed.
• Amend Hernan to Hernán.
• Use uv to control Python in the GHA CI pytest tests.