v4.3

This version fixes all remaining issues left after the refactor done between v4.0 and v4.1. It also completes the entire documentation, and makes the package citable.

This version should be considered stable and ready to use for end-users.

v4.3.0-rc1

What's Changed

This version fixes all remaining issues left after the refactor of v4.1. It also completes the entire documentation.

Once fully released, this version should be consider stable and ready for end-users.

Details

• fixes threshold method, order of variables gamma and lambda was reversed by @guillermoaguilar in #139
• Fix that plot_psychometric_function draws gamma incorrectly by @pberkes in #148
• Demo notebook for using psignifit in a log-space by @pberkes in #149
• Improve sigmoid tests, fix bug in Gaussian._slope by @pberkes in #146
• Remove estimate type parameter by @pberkes in #152
• Return the bounds dictionary in debug mode by @guillermoaguilar in #153
• Fix plot_prior to plot 25 and 75 percentile of prior by @pberkes in #154
• Change the infrastucture of the docs to jupyterbook by @lschwetlick in #144
• docs: adds warning on all pages again with proper syntax, also on generated examples by @guillermoaguilar in #155
• docs: make build docs also on pull request by @guillermoaguilar in #157
• docs: adds banner on top of book. by @guillermoaguilar in #156
• correct plot_bayes [WIP] by @guillermoaguilar in #159
• Save confidence intervals as a dictionary by @pberkes in #163
• docs: Migrate old rst files (sphinx) to markdown format by @guillermoaguilar in #165
• docs: improve demo where we plot all sigmoids by @guillermoaguilar in #167
• docs: improve basic usage demo, now executable. by @guillermoaguilar in #166
• Add the "posterior mean" parameter estimate type by @pberkes in #161
• Implement a Sigmoid.standard_parameters method by @pberkes in #169
• docs: correcting documentation by @guillermoaguilar in #168
• Remove obsolete configuration options by @otizonaizit in #171
• docs: renames parameter_estimate to get_parameters_estimate() by @guillermoaguilar in #174
• Fix computation of confidence interval Result.threshold method by @pberkes in #178
• Reinstate Result.parameter_estimate by @otizonaizit in #179
• Exit with an error when building docs if a notebook fails to execute by @otizonaizit in #181
• Fixes plot in interval bias analysis by @guillermoaguilar in #182
• Fixes wrong position of black dot in prior plot by @guillermoaguilar in #186
• Adds 2D marginals also for fixed params by @guillermoaguilar in #187
• Change mentions of "p-levels" by @pberkes in #188
• Move rescaling of psychometric function related to lambda and gamma in Sigmoid object by @pberkes in #184
• Move tests and demos out of psignifit package by @pberkes in #185
• Remove tools.psychometric function by @pberkes in #190
• Updates documentation - Priors by @guillermoaguilar in #192
• Parametrize tests using the sigmoid classes, not the names by @pberkes in #193
• remove obsolete file by @otizonaizit in #197
• Posterior samples by @pberkes in #196
• Adds demos to the docs as md files by @lschwetlick in #198
• Restore stimulus range as a public configuration variable by @otizonaizit in #201
• Add a docs page for the get_standard_parameters method. by @lschwetlick in #202
• Docstrings for the functions in psignifit and psigniplot by @pberkes in #203
• Update Docs: Pooling by @lschwetlick in #199
• Fix alpha_width config by @pberkes in #205
• Update Docs: advanced options page by @lschwetlick in #200
• Revise docstrings in sigmoids.py and tools.py by @pberkes in #206
• Fixes formatting of warning in documentation for advanced options by @guillermoaguilar in #207
• Allow setting thresh_PC to something other than 0.5 by @pberkes in #208
• Fixes block number plotting in psigniplot model fit by @guillermoaguilar in #209
• update contributor list by @otizonaizit in #210
• cast estimated parameters from Numpy scalars to Python floats by @otizonaizit in #212
• Reformulate docs by @otizonaizit in #211
• build docs using oldest supported Python version and add Python 3.13 to CI tests and release workflows by @otizonaizit in #215
• Add a proportion_correct method to the Result object to get proportion correct values from the fitted psychometric function by @otizonaizit in #214
• get rid of obsolete file and update copyright date by @otizonaizit in #216
• Add to the matrix Python==3.12, numpy==1.26.4 by @pberkes in #224
• Fixes timeout issue happening sometimes when building the docs by @guillermoaguilar in #230
• Fixed parms tests by @otizonaizit in #225
• Hist docs by @otizonaizit in #226
• Document where the name psignifit comes from by @otizonaizit in #231
• Add couple of more datasets to test against by @guillermoaguilar in #233
• Adds in docs basic usage a visualization of the psychometric function parameters by @guillermoaguilar in #232

Full Changelog: v4.2...v4.3.0-rc1

v4.2

What's Changed

• Update contributors list #129
• Make test for eta parameter recovery more stable in the face of randomness #131
• Make plot aesthetics equal to matlab #134
• Set default CI method to "percentile" and add a test #135
• Silence warnings in tests #132
• Fixes all plot functions #138

Full Changelog: v4.1...v4.2

v4.1

Complete refactor of previous version 4.0.0.

v4.0.0

First version of python psignifit.

This version was a direct translation from MATLAB, not pythonic and not optimized to run with python libraries.
Posterior versions (v.4.1 and onwards) are not backwards compatible to this version; this is because a big refactoring effort was needed in order to bring the library to python standards.

Thus this release serves only for legacy purposes: in case you have code that used python-psignifit v4.0 and still want to be able to run it. Still, we strongly recommend you to upgrade your code to use the syntax of the newest release.