Enhancements:
- Allow entry of Type A measurement data directly into Uncertainty interface
- Updated cost-based guardbanding to work with arbitrary distributions
- (Re)add Level of Confidence vs Coverage Factor dropdown in expanded uncertainty results
- Simplified data importing dialogs
- Moved some Risk calculation functions to Tools menu
- Added PFR Target guardbanding
- Show worst-case specific risk in risk calculation report
- Enabled dark mode switching
Bug Fixes:
- Fixed PFR calculation with single-sided limits
- Fixed guardbanding in Risk Curves tool
- Fixed potential issue calculating uncertainty with expression containing log of constant value
Other updates:
- Changed DataSets API to match other calculation types
- Removed deprecated UncertCalc class
- Upgraded UI to use QT6
- Updated for compatibility with Python 3.12
- Added inline help documentation
- Risk Analysis: Added calculation of conditional false accept risk and guardbanding based on conditional risk
- Risk Analysis: Added option to set process distribution using ITP
- Data Sets: Fixed errors when saving and displaying histograms with empty data
- Other minor bug fixes
API Changes - Risk Analysis (possibly breaking):
- Target PFA guardband calculation
risk.guardbandis nowrisk.guardband.target. GBparameter ofPFA_normandPFR_normmethods must be float. To calculate GB using a defined method, use methods inrisk.guardband_tur.*.- Removed
approxkeyword from riskPFA,PFRandguardband.target methods. PFAR_MCreturns an additional element for Conditional PFA.
- Added re() and im() functions for ModelComplex calculations
- Fixed evaluation of constant model functions
- Corrected possible skew in histogram of Monte Carlo Distribution output plot
- Fixed issues with offset (eg temperature) unit conversion in uncertainty values
- Fixed issue when importing a distribution with measurement units
- Fixed entry of uncertainty level of confidence
- Replaced setup.py with setup.cfg
- Improve convergence of S2 Interval Calculation minimization problem
- Enabled coth and acoth functions
- Allow entering constant quantities with units into measurement model
- Restored sensitivity coefficients in GUM Derivation report
- Fixed GUI issues, including entering of chained uncertainty models, and reverse function names
- Fixed some reports when Monte Carlo results contained NaN or Inf values
UI changes:
- Implemented guided uncertainty wizard interface
- Added resizable splitters between sections in the window
- Updated ANOVA reports to better align with GUM examples
API changes (possibly breaking):
- Refactored API to use Model and ModelCallable classes, and deprecated UncertCalc class.
- Updated return values into Results objects and Reports objects for consistency across all calculation types.
- Updated for compatibility with Python 3.10 and 3.11. Minimum required version is 3.9.
- Makes some imports from subpackages
- Added API User Manual as Jupyter Notebooks in docs folder
- Improved plot and equation resolution when using high-DPI displays
- Fixed importing CSV files with single column
- Fixed accept/reject indication on risk calculation with asymmetric guardbands
- Added configuration for Curve Fit full report
- Corrected off-by-one error in autocorrelation calculation
- Handle several edge-case errors
- Fixed Jupyter representation of plots
- Allow up to 8 significant figures in reports
- Account for process bias when computing decision risk from in-tolerance probability
- Fixed reporting of Monte Carlo sensitivity coefficients for non-vectorizable functions
- Fixed risk plotting when one specification limit is infinite, or invalid guardband factor is entered
- Restored axis labels in risk distribution plots
- Corrected the equation displayed for prediction band curve fit uncertainty based on u(y) mode
- Added joint Risk probability plot
- Added YAML filter to file open/save dialog
- Improved some report formatting
- Fixed importing of data into summarized interval table
- Prevent exception on corrupted settings file
- Prevent exception when no name is given in distribution explorer
- Restore warning when functions contain circular references
- Improve moving array data from DataSet into sweeps or curve fits
- Fix issues when importing single-asset data into interval calculations
- Show Deviation from Prior plot in interval variables method report
- Improved automatic conversion of delta units
- Restore saving full report in curve fit calculation
- Allowed units checker window to show message for each measurement function
- Fixed save/load of dataset calculation in summary statistics mode
- Fixed entering of correlation sweep ranges
- Added more summary statistics to datasets histogram page
- Fixed entering of curvtrap distribution parameters
- Improvements to CSV data importing
- Move PyQt5 to an optional dependency, only required for GUI
- Fixed Mac builds
- Updated some Markdown tables for correct formatting
- Added custom parameterizations of Exponential, Beta, and Lognorm distributions
- Select between "mean" and "median" in for distribution entry
- Changed gamma distribution to input alpha and beta parameters, consistent with JCGM.
- Added "observed itp" parameter to PFA_norm and PFR_norm functions
- Fixed scrolling issue on Windows
- Fixed issue when a measurement function is a constant
- Updates for Python 3.8+ and dependency compatibility
- Improved uncertainty computation and derivation report in multi-output measurement models
- Improved handling of callable functions with multiple outputs
- Added colorbar to contour plots
- Allow UncertComplex to compute uncertainty on callable functions
- Suppress duplicate plot outputs in Jupyter
- Fixed order of polynomial curve fit coefficients in report
- Fixed date issue introduced by Matplotlib 3.3
- Fixed font scaling issue in Windows with display scaling enabled
- Added calibration interval calculator, using NCSLI RP-1 Method A3 and Method S2
- Implemented "Symbolic GUM Only" mode to show GUM equation results without entering values
- Added risk sweeps mode for plotting PFA(R) vs. itp at various TUR, or other combinations
- Report the correlation matrix between multiple uncertainty model equations
- Save user-defined units with project files
- Added distribution fit selection to Data Sets page
- Added random seed option to Distribution Explorer
- Sort the measured variables list alphabetically
- Fixed issue entering t-distribution parameters before degrees of freedom is defined
- Fixed display of degree symbol in output reports
- Handle callable measurement models that raise exception with Unit Quantity inputs
- Minor fixes for Python 3.8 compatibility
- Fixed potential units conversion error in Welch-Satterthwaite degrees of freedom calculation
- Repeatability and reproducibility, as standard as standard deviation of the mean, can be imported from datasets
- Added guardband sweep, probability of conformance plots to risk calculator
- Finished flattened uncertainty entry user interface
- Fixed error in computing Poisson distribution mean for Monte Carlo calculation
- Fix a possible unit conversion error
- Improved output reporting system. Adds right-click option to output reports for changing significant figures and number formats.
- Flattened uncertainty entry user interface and improved keyboard navigation
- GUI enhancements and fixes
- GUI enhancements and fixes
- Added autocorrelation calculation to Data Sets and ANOVA
- Improved user interface for data sharing between calculation types
- Show confidence band on normal probability plots
- Fix exception due to units incompatibility when calculating uncertainty of k**x
- Fix units conversion issue with Monte Carlo of non-vectorizable Python functions
- Minor bug fixes
- Added processing and conversion of measurement units
- Added Monte-Carlo option for computing risk probabilities
- Other minor user interface enhancements
Initial Public Release