Reliability

'''''''''''''''''''''''''''''''''''''''''''

**Summary of changes**

This is a major release. After 14 months of no updates, this release is essential to fix issues with dependency compatability, mainly with scipy. More frequent updates are planned in the near future.

**Bug Fixes**

- Issues `51 <https://github.com/MatthewReid854/reliability/issues/51>`_ and `54 <https://github.com/MatthewReid854/reliability/issues/54>`_ have been resolved by updating "simps" to "simpson" from scipy.
- Issue `55 <https://github.com/MatthewReid854/reliability/issues/55>`_ has been resolved. There was a bug that occurred in the plots for Fitters.Fit_Everything when all the failure times were small numbers (~0.01).
- Issue `49 <https://github.com/MatthewReid854/reliability/issues/49>`_ has been resolved. There was a bug with Mixture, Competing Risks, and DSZI models for the quantile and inverse_SF methods that did not accept array like inputs. This was resolved by changing the method used to find these values to interp.

**Other**

- Requirements for all the dependencies have been updated to the most recent releases.
- Tests have been updated to reflect the slightly more accurate results when issue 49 was fixed.

'''''''''''''''''''''''''''''''''''''''''''''

**Summary of changes**

This is a minor release to fix two bugs. The first bug was in plot legend for PoF.SN_diagram which had the confidence intervals hardcoded to 95%. The legend now reflects the specified CI.
The second bug was from `this issue <https://github.com/MatthewReid854/reliability/issues/46>`_ where the Lognormal Distribution CDF didn't return a tuple. This is now fixed and aligns with the remainder of the distributions.
A few broken hyperlinks in the docs have also been fixed.

'''''''''''''''''''''''''''''''''''''''''''''

**Summary of changes**

This is a minor release to add the ability to use Mixture_Model and Competing_Risks_Model as distributions in Other_functions.stress_strength.
There is also a minor bugfix to allow Loglogistic distributions with beta<1 to be used. The bug occurred because a text string was created for the mean of a Loglogistic distribution with beta<1 (since there is no mean for such distributions) and the mean was used for a calculation causing an error. The calculation works equally well using the median which does exist for these distributions.
Thanks to Cole Tenold for raising this `issue <https://github.com/MatthewReid854/reliability/issues/45>`_.

'''''''''''''''''''''''''''''''''''''''''''''

**Summary of changes**

This is a minor release to fix a bug in Other_functions.stress_strength and Other_functions.stress_strength_normal. The bug occurred when stress was greater than strength causing the integration to fail and return incorrect values for the probability of failure. This has been fixed and the plots are also improved. Thanks to github user ctenold for identifying `this issue <https://github.com/MatthewReid854/reliability/issues/43>`_ and providing a working solution.
There is also a speed improvement for Fitters.Fit_Weibull_2P_grouped which takes advantage of the np.repeat function to optimize the process of generating the arrays for large datasets of repeating values. Thanks to Tomas Santos for identifying this improvement.

'''''''''''''''''''''''''''''''''''''''''''''

**Summary of changes**

This is a minor release to fix a bug in the plot of Reliability_testing.chi2test identified in `this issue <https://github.com/MatthewReid854/reliability/issues/41>`_. The normed parameter in numpy.histogram is deprecated and had been replaced by the density parameter.

'''''''''''''''''''''''''''''''''''''''''''''

**Summary of changes**

This is a minor release to fix a bug in the plot of Nonparametric.RankAdjustment identified in `this issue <https://github.com/MatthewReid854/reliability/issues/40>`_. The values weren't being sorted. This has been fixed by adding an optional 'sort' argument to Probability_plotting.plotting_positions.