From 43049d3c8fd55f51952e5354ccaa4997383442d0 Mon Sep 17 00:00:00 2001 From: ejike ugba Date: Mon, 25 Oct 2021 15:26:04 +0200 Subject: [PATCH] Version 0.2.2 --- DESCRIPTION | 2 +- JOSS/paper.md | 3 --- NEWS.md | 40 ++++++++++++++++++++++++---------------- 3 files changed, 25 insertions(+), 20 deletions(-) diff --git a/DESCRIPTION b/DESCRIPTION index a5839dd..c85f690 100644 --- a/DESCRIPTION +++ b/DESCRIPTION @@ -1,7 +1,7 @@ Type: Package Package: serp Title: Smooth Effects on Response Penalty for 'CLM' -Version: 0.2.1 +Version: 0.2.2 Authors@R: person(given = "Ejike R.", family = "Ugba", diff --git a/JOSS/paper.md b/JOSS/paper.md index 5ae5d52..93cb611 100644 --- a/JOSS/paper.md +++ b/JOSS/paper.md @@ -25,9 +25,6 @@ bibliography: paper.bib The use of specialized methods for the analysis of categorical data has been on the rise in recent years [@agresti_analysis_2010]. For instance, scientists frequently use the different forms of the ordinal model to analyze relationships between an ordinal response variable and covariates of interest. In particular, the cumulative link model (CLM) propagated by @mcCullagh_regression_1980 finds a wide range of empirical applications in clinical trials, social surveys, market research, etc. However, in high-dimensional settings with a large number of unknown parameters, e.g., if several potential predictors and response categories are modeled, the so-called identification problem could be somewhat unavoidable [@fisher_identification_1966; @bartels_identification_1985]. Thankfully, regularization techniques [see, e.g., @bhlmann_statistics_2011;@hastie_2009] among other approaches, offer a remedy in such situations. -The use of specialized methods for the analysis of categorical data has been on the rise in recent years [@agresti_analysis_2010]. For instance, scientists frequently use the different forms of the ordinal model to analyze relationships between an ordinal response variable and covariates of interest. In particular, the cumulative link model (CLM) propagated by @mcCullagh_regression_1980 finds a wide range of empirical applications in clinical trials, social surveys, market research, etc. However, in high-dimensional settings with a large number of unknown parameters, e.g., if several potential predictors and response categories are modeled, the so-called identification problem could be somewhat unavoidable [@fisher_identification_1966; @bartels_identification_1985]. Thankfully, regularization techniques [see, e.g., @hastie_2009;@bhlmann_statistics_2011] among other approaches, offer a remedy in such situations. ->>>>>>> d5f5699 (Remove extra spaces) - The `serp` software package [@Ugba_serp_2021], available in the Comprehensive R Archive Network ([CRAN](https://CRAN.R-project.org/package=serp)) [@R_language_2021], implements a unique regularization algorithm that enforces smoothness on the adjacent categories of CLMs. The approach provides flexible modeling of CLMs that uses the smooth-effects-on-response penalty (SERP) discussed in @ugba_smoothing_2021 and @tutz_regularized_2016. For instance, when applied to the general form of the cumulative logit model, also known as the non-proportional odds model (NPOM), one obtains the proportional odds model (POM) under extreme parameter shrinkage. Fitting in `serp` is carried out by a modified Newton's optimization method that facilitates an easy convergence and estimation speed. As an open-source software package, `serp` aims to provide a platform for advanced scientific modeling in empirical research. The core features of `serp`, as well as details about usage, are provided in the software package [documentation](https://cran.r-project.org/package=serp/serp.pdf). diff --git a/NEWS.md b/NEWS.md index c0bd252..1ef1961 100644 --- a/NEWS.md +++ b/NEWS.md @@ -1,15 +1,23 @@ -# serp 0.2.1 +## serp 0.2.2 +- JOSS release +- minor changes in serp documentation +- provide coefficient() as an alias for coef() +- update README.md to include community guidelines and contributors code of conduct +- + +--- +## serp 0.2.1 - submission to CRAN --- -# serp 0.2.0.9001 +## serp 0.2.0.9001 - update function included in namespace - examples included in the different function documentation - shrinkage parameter upper limit set to 1e10 in serp.control - bug fix in test function --- -# serp 0.2.0.9000 +## serp 0.2.0.9000 - deviance tuning option in serp tuneMethod now replaced by AIC - serp output includes residual degrees of freedom (rdf) - function to compute the effective degrees of freedom and rdf from the trace of the generalized hat matrix provided @@ -18,46 +26,46 @@ - changes made in README.Rmd and README.md --- -# serp 0.2.0 +## serp 0.2.0 - serp version 0.2.0 release --- -# serp 0.1.9.9001 +## serp 0.1.9.9001 - re-submission to CRAN --- -# serp 0.1.9.9000 +## serp 0.1.9.9000 - fixed all issues spotted out in the last version released on CRAN. --- -# serp 0.1.9 +## serp 0.1.9 - Submit serp version 0.1.9 to CRAN --- -# serp 0.1.8.9007 +## serp 0.1.8.9007 - updates README.md - description gets additional doi - references in serp documentation updated --- -# serp 0.1.8.9006 +## serp 0.1.8.9006 - Bugs in tests fixed --- -# serp 0.1.8.9005 +## serp 0.1.8.9005 - tests reconstructed to yield improved unit test coverage. - rewrote some of the error and warning messages in key serp functions. - Bugs in anova.serp fixed. - --- -# serp 0.1.8.9004 +## serp 0.1.8.9004 - update License in the description - Bugs in the example codes fixed - Bugs in test files corrected --- -# serp 0.1.8.9003 +## serp 0.1.8.9003 - Column titles of predicted values with reverse TRUE are corrected. - Bugs in the deviance (dvfun) are corrected. - serp predict function now handles both single and multiple row input(s). @@ -72,7 +80,7 @@ - predicted values in errorMetrics get normalized for greater efficiency. --- -# serp 0.1.8.9002 +## serp 0.1.8.9002 - warning messages in serpfit get updated. - Bug in serpfit when specifying gridType corrected. - trainError no longer shows up in serp output. @@ -86,15 +94,15 @@ - updates in serp.control warning messages. --- -# serp 0.1.8.9001 +## serp 0.1.8.9001 - serp gets a new license. --- -# serp 0.1.8.9000 +## serp 0.1.8.9000 * README.md gets additional badges and a hexagon sticker --- -# serp 0.1.8 +## serp 0.1.8 * package re-submission to CRAN - reference to methods used in the package is included in the description field