alg-ml >> {PPtreExt}; diag-reg >> {snreg}; fcast-nonlin >> {giancarlo-verse}; phoc-gen >> {bayesics}

Author: ercbk

qmd/algorithms-ml.qmd

@@ -327,14 +327,16 @@
     -   Tessellations partition the covariate space and output values are based on the cell the sample falls in.
         -   Applies a probabilistic approach to which cell a sample falls into based on the distance to the center of its cell and the others then this would further smooth the output function of the algorithm (similar to SoftBART).
 -   [{]{style="color: #990000"}[bonsai](https://bonsai.tidymodels.org/){style="color: #990000"}[}]{style="color: #990000"} - tidymodels extension fits [{partykit::ctree}]{style="color: #990000"}
--   [{]{style="color: #990000"}[partykit](https://cran.r-project.org/web/packages/partykit/index.html){style="color: #990000"}[}]{style="color: #990000"} - model-based trees
 -   [{]{style="color: #990000"}[glmertree](https://cran.r-project.org/web/packages/glmertree/index.html){style="color: #990000"}[}]{style="color: #990000"} - Generalized Linear Mixed Model Trees
     -   Combines `lmer`/`glmer` from [{lme4}]{style="color: #990000"} and `lmtree`/`glmtree` from [{partykit}]{style="color: #990000"}
 -   [{]{style="color: #990000"}[LDATree](https://iamwangsiyu.com/LDATree/){style="color: #990000"}[}]{style="color: #990000"} ([Paper](https://arxiv.org/abs/2410.23147)) - Integrates Uncorrelated Linear Discriminant Analysis (ULDA) and Forward ULDA into a decision tree structure
     -   Superior to traditional decision trees and other implementations (see paper) of oblique trees.
     -   Effectively generates oblique splits, handles missing values, performs feature selection, and outputs predicted class labels and class probabilities.
     -   Peforms extremely well when the dataset contains many noise variables, and when significant high-order interactions are present alongside non-significant low-order interactions. In these scenarios, the proposed method outperforms other methods, including the random forest.
 -   [{]{style="color: #990000"}[parttree](https://grantmcdermott.com/parttree/){style="color: #990000"}[}]{style="color: #990000"} - Visualize simple 2-D decision tree partitions
+-   [{]{style="color: #990000"}[partykit](https://cran.r-project.org/web/packages/partykit/index.html){style="color: #990000"}[}]{style="color: #990000"} - model-based trees
+-   [{]{style="color: #990000"}[PPtreeExt](https://natydasilva.github.io/PPtreeExt/){style="color: #990000"}[}]{style="color: #990000"} - Extends to the Projection Pursuit Tree (PPtree) algorithm to improve its performance in multi-class settings and under nonlinear separations
+    -   The PPtree classifier finds separations between classes based on linear combinations of variables by optimizing a projection pursuit index.
 -   [BART]{.underline}
     -   [{]{style="color: #990000"}[bartMan](https://alaninglis.github.io/bartMan/){style="color: #990000"}[}]{style="color: #990000"} - Investigating and visualizing Bayesian Additive Regression Tree (BART) model fits.
         -   Currently supports [{]{style="color: #990000"}[BART](https://cran.r-project.org/web/packages/BART/index.html){style="color: #990000"}[}]{style="color: #990000"}, [{]{style="color: #990000"}[dbarts](https://cran.r-project.org/web/packages/dbarts/index.html){style="color: #990000"}[}]{style="color: #990000"}, and [{bartMachine}]{style="color: #990000"}

qmd/diagnostics-regression.qmd

@@ -219,6 +219,9 @@ fig-cap-location: top
             -   Weighted Least Squares (see [Regression, Other \>\> Weighted Least Squares](regression-other.html#sec-reg-other-wls){style="color: green"})
                 -   Also see Feasible Generalized Least Squares (FGLS) in the same note
                 -   Example: [Real Estate \>\> Appraisal Methods \>\> CMA \>\> Market Price](https://ercbk.github.io/Domain-Knowledge-Notebook/qmd/real-estate.html#sec-rlest-apprais-ml-cmaprc-marprc){style="color: green"} \>\> Case-Shiller method
+            -   [{]{style="color: #990000"}[snreg](https://cran.r-project.org/web/packages/snreg/index.html){style="color: #990000"}[}]{style="color: #990000"} - Regression with Skew-Normally Distributed Error Term
+                -   Estimate regression models with skew‑normal error terms, allowing both the variance and skewness parameters to be heteroskedastic.
+                -   Includes a stochastic frontier framework that accommodates both i.i.d. and heteroskedastic inefficiency terms.
             -   Scale model (`greybox::sm` ) models the variance of the residuals or `greybox::alm` will call `sm` and fit a model with estimated residual variance
                 -   See [article](https://forecasting.svetunkov.ru/en/2022/01/23/introducing-scale-model-in-greybox/) for an example
                 -   Can be used with other distributions besides gaussian

qmd/forecasting-nonlinear.qmd

@@ -3,8 +3,6 @@
 ## Misc {#sec-fcast-nlin-misc .unnumbered}
 
 -   Gaps in the time variable can be a problem if you are trying to interpolate between those gaps. (see bkmk, `method = "reml" + s(x, m = 1)`)
--   Resources
-    -   Nonlinear Time Series Analysis, Kantz, Schreiber (See R \>\> Documents \>\> Time Series)
 -   Packages
     -   Time Series [Task View](http://cran.r-project.org/web/views/TimeSeries.html) "Nonlinear Time Series Analysis"
     -   [{]{style="color: #990000"}[aplms](https://cran.r-project.org/web/packages/aplms/index.html){style="color: #990000"}[}]{style="color: #990000"} - Additive Partial Linear Models with Symmetric Autoregressive Errors
@@ -16,6 +14,8 @@
     -   [{]{style="color: #990000"}[EstemPMM](https://cran.r-project.org/web/packages/EstemPMM/index.html){style="color: #990000"}[}]{style="color: #990000"} - Implements the Polynomial Maximization Method ('PMM') for parameter estimation in linear and time series models when error distributions deviate from normality.
         -   The 'PMM2' variant achieves lower variance parameter estimates compared to ordinary least squares ('OLS') when errors exhibit significant skewness.
         -   Includes methods for linear regression, 'AR'/'MA'/'ARMA'/'ARIMA' models, and bootstrap inference.
+    -   [{]{style="color: #990000"}[giancarlo-verse](https://rpubs.com/giancarlo_vercellino){style="color: #990000"}[}]{style="color: #990000"} - Giancarlo is building a lot of wonderfully complex model packages, so I'm just going to link his R-Pubs page which has intros to these packages.
+        -   [{]{style="color: #990000"}[wired](https://cran.r-project.org/web/packages/wired/index.html){style="color: #990000"}[}]{style="color: #990000"} - Weighted Adaptive Prediction with Structured Dependence
     -   [{]{style="color: #990000"}[nonlinearTseries](https://cran.r-project.org/web/packages/nonlinearTseries/){style="color: #990000"}[}]{style="color: #990000"} ([Vignette](https://cran.r-project.org/web/packages/nonlinearTseries/vignettes/nonlinearTseries_quickstart.html)) - Facilitates the computation of the most-used nonlinear statistics/algorithms including generalized correlation dimension, information dimension, largest Lyapunov exponent, sample entropy and Recurrence Quantification Analysis (RQA), among others. Basic routines for surrogate data testing are also included.
     -   [{]{style="color: #990000"}[probcast](https://github.com/jbrowell/ProbCast){style="color: #990000"}[}]{style="color: #990000"} - Has function wrappers around gams, gamlss, and boosted gamlss models from {mgcv}, {mboost}, {gamlss}, etc. for use in forecasting. Supports high-dimensional dependency modeling based on Gaussian Copulas ([paper](http://www.jethrobrowell.com/uploads/4/5/4/0/45405281/probcast___pmaps2020.pdf), [use case](https://forecasting.svetunkov.ru/en/2023/05/09/probabilistic-forecasting-of-hourly-emergency-department-arrivals/))
     -   [{]{style="color: #990000"}[tsDyn](https://cran.r-project.org/web/packages/tsDyn/index.html){style="color: #990000"}[}]{style="color: #990000"}
@@ -26,6 +26,7 @@
         -   AAR: additive AR
 -   Resources
     -   [Temporal autocorrelation in GAMs and the mvgam package](https://ecogambler.netlify.app/blog/autocorrelated-gams/)
+    -   Nonlinear Time Series Analysis, Kantz, Schreiber (See R \>\> Documents \>\> Time Series)
 -   Copulas
     -   Also see [Association, Copulas](association-copulas.qmd#sec-assoc-cop){style="color: green"}
     -   [TUTORIAL julia, copulas + ARMA model, example w/exonential distribution - ARMA Forecasting for Non-Gaussian Time-Series Data Using Copulas \| by Sarem Seitz \| Jun, 2022 \| Towards Data Science](https://towardsdatascience.com/arma-forecasting-for-non-gaussian-time-series-data-using-copulas-45a3a28f69e5)

qmd/post-hoc-analysis-general.qmd

@@ -19,6 +19,9 @@
 
 -   Packages
 
+    -   [{]{style="color: #990000"}[bayesics](https://cran.r-project.org/web/packages/bayesics/index.html){style="color: #990000"}[}]{style="color: #990000"} - Perform fundamental analyses using Bayesian parametric and non-parametric inference (regression, anova, 1 and 2 sample inference, non-parametric tests, etc.).
+        -   (Practically) no MCMC is used; all exact finite sample inference is completed via closed form solutions or else through posterior sampling automated to ensure precision in interval estimate bounds.
+        -   Diagnostic plots for model assessment, and key inferential quantities (point and interval estimates, probability of direction, region of practical equivalence, and Bayes factors) and model visualizations are provided.
     -   [{]{style="color: #990000"}[dabestr](https://acclab.github.io/dabestr/index.html){style="color: #990000"}[}]{style="color: #990000"} for visualization
     -   [{]{style="color: #990000"}[inferno](https://github.com/pglpm/inferno/tree/main){style="color: #990000"}[}]{style="color: #990000"} - Uses MCMC and Bayes to estimate frequency proportions and mutual information calculations.
         -   The method is referred to as Bayesian nonparametric population inference, which can also be called "inference under exchangeability" or "density inference". I'm not sure exactly what that means.