nlsCompare

An R package to compare different packages' functions for nonlinear least squares (nls)

This package grew out as a product of the Google Summer of Code (2021) project Improvements to nls() by Arkajyoti Bhattacharjee and mentored by Dr. John C Nash and Dr. Heather Turner.

Current Workflow

Get a one-line machine summary using machineId().
Select directory to save outputs of program using setup_dir().
Create dataframe to store the database of problems, algorithms and controls associated with nls() like functions using create_db().
Create dataframe to store the error log. This shows which "solver" fails in "which problem" and "why?". This shows limitations, improvements and areas of potential imporovements. Use create_elog().
Check if the final output csv-s - "nlsDatabase.csv" and "nlsErrorLog.csv" exists in the directory chosen in step 2.
Run the main program - comparison-testing and corresponding editing of the dataframes - using run().
Write the dataframes into "nlsDatabase.csv" and "nlsErrorLog.csv" using write_csvs().
Finally, remove the global variables created in the above steps using rm_nls().

Documentation

A vignette to demonstrate the above workflow is nlsCompare: How to use it?.

For more detals about the package, please refer to nlsCompareArticle.

Installation

Warning: As of 2021-08-23, the package is still in the development stage!

To install this development repo via devtools in R, use:

# install.packages("devtools")
devtools::install_github("ArkaB-DS/nlsCompare")