Changelogs » Psytrack
PyUp Safety actively tracks 281,944 Python packages for vulnerabilities and notifies you when to upgrade.
Psytrack
2.0.1
- Update to README to reflect publication in Neuron.
2.0.0
- Primarily superficial changes ahead of the release of final paper.
1.3.0
- Added a `hess_calc` argument to `recoverSim` to control whether error bars are calculated in the recovery of simulated weights. - Added a `hess_calc` argument to `crossValidate` to control whether error bars are calculated in the recovery of simulated weights. - Change display of error bars from 2 standard errors, to 1.96 which maps more precisely to a 95% credible interval. - Update the analysis function to handle `s_a` and `s_b` as inputs. - Update the analysis function to better plot session boundaries. - Update the analysis function to handle Left/Right sides coded as {0,1}, as well as {1,2}. - The default z-order (plotting order) of different inputs can now be accessed through `psy.ZORDER`, similar to `psy.COLORS`. - Made a few superficial updates to the tutorial notebook.
1.2.0
- This may not be a backward compatible update, apologies in advance! - Added functionality to `hyperOpt` for returning error bars for the weights returned in `best_wMode` by setting the keyword argument `hess_calc`. - Added functionality to `hyperOpt` for returning error bars for the hyperparameters returned in `best_hyper` by setting the keyword argument `hess_calc`. - Added check to `getMap` raise error if choices `y` are not formatted correctly. - Corrected a minor zero-indexing bug in the cross-validation code. - Adjusted `runSim` code to simulate and recover models using the `sigDay` functionality. - Update documentation to clarify that the order of weights returned in `wMode` is alphabetical with respect to the dictionary `weights`. - Adjust the `__init__.py` file to make commonly used functions more directly accessible. - Large improvements in the plotting functions, including more modularity and configurability. - Made cross-validation much more straightforward, should be able to get log-likelihood direct from a single function now. - Refresh of tutorial notebook
1.1.0
- Rename `aux` folder to `helper` to allow for installation in Windows systems