FoKL 3.2.0

• kernel using orthogonal Bernoulli polynomials as basis functions is now included (primarily for Pyomo application)
• some tutorials added
• README extensively updated/overhauled with more thorough documentation
• several minor bug fixes / patch updates
• original research paper included as PDF
• structure of 'clean' method modified slightly to allow auto normalization and formatting of inputs without data
• option to pass dataset into 'clean' method from within argument of 'fit' and 'evaluate' methods
• folder '_dev' added to begin collecting past and present useful materials for development to have for future reference
• list of dependencies added

FoKL 3.1.1

• patched 'getKernels' method of getting kernel txt file location to work on non-Windows OS

FoKL 3.1.0

• [MAJOR]: fixed bug where 'killset' was eliminating the wrong terms due to an indexing error caused by 'killset' not always being ordered low to high, so 'mtx' now aligns with MATLAB for larger datasets with more than one eliminated term
• created 'inputs_to_phind' function to eliminate duplicated code (function is called 3 times), using phind method that was in bss_derivatives (aligns with MATLAB results)
• added 'main()' guard to Sigmoid and GP integrate examples
• minor changes like indentations on long lines of code / comments / documentation

FoKL 3.0.3

Bug Fixes in BSS Derivatives

• Code is not treating terms that do not contain the variable being differentiated by appropriately.

FoKL 3.0.2

• second derivatives added to bss_derivatives
• README updated to reflect second derivatives
• getKernels now reads from text file and has functionality to smooth raw coefficients and save as a new file
• folder added to src/FoKL for housing various kernel text files (i.e., spline coefficients)

FoKL 3.0.1

• Additional function 'bss_derivatives' for taking partial first derivatives (i.e., gradient) of model with respect to each input variable

FoKL 3.0.0

• Initialize model with only user-defined hyperparameters, with the rest set to default values
• Automatic formatting of inputs and data, including normalization of inputs
• Keyword 'train' for automatic test/train splits
• Keywords added to model.coverage3() for user control of plotting
• Storage of various info such as inputs, data, normalization scale factors, etc. as attributes of model
• Function model.clear() for removing attributes from model that are not hyperparameters, to enable sweeping / iterating through datasets without needing to create a new FoKL class

FoKL 2.0.1

• Documentation update
• Updated example scripts to match new functionality

FoKL 2.0.0

• Changes in how FoKL is called
• Updated examples to match MatLab repo
• Fixed issues in GP_Intergrate

2.0 beta

File structure changes pre-release