sns - Stochastic Newton Sampler (SNS)

Stochastic Newton Sampler (SNS) is a Metropolis-Hastings-based, Markov Chain Monte Carlo sampler for twice differentiable, log-concave probability density functions (PDFs) where the proposal density function is a multivariate Gaussian resulting from a second-order Taylor-series expansion of log-density around the current point. The mean of the Gaussian proposal is the full Newton-Raphson step from the current point. A Boolean flag allows for switching from SNS to Newton-Raphson optimization (by choosing the mean of proposal function as next point). This can be used during burn-in to get close to the mode of the PDF (which is unique due to concavity). For high-dimensional densities, mixing can be improved via 'state space partitioning' strategy, in which SNS is applied to disjoint subsets of state space, wrapped in a Gibbs cycle. Numerical differentiation is available when analytical expressions for gradient and Hessian are not available. Facilities for validation and numerical differentiation of log-density are provided. Note: Formerly available versions of the MfUSampler can be obtained from the archive <https://cran.r-project.org/src/contrib/Archive/MfUSampler/>.

Last updated 2 years ago

3.30 score 1 dependents 33 scripts 204 downloads

MfUSampler - Multivariate-from-Univariate (MfU) MCMC Sampler

Convenience functions for multivariate MCMC using univariate samplers including: slice sampler with stepout and shrinkage (Neal (2003) <DOI:10.1214/aos/1056562461>), adaptive rejection sampler (Gilks and Wild (1992) <DOI:10.2307/2347565>), adaptive rejection Metropolis (Gilks et al (1995) <DOI:10.2307/2986138>), and univariate Metropolis with Gaussian proposal.

Last updated 2 years ago

3.08 score 2 dependents 20 scripts 319 downloads

RegressionFactory - Expander Functions for Generating Full Gradient and Hessian from Single-Slot and Multi-Slot Base Distributions

The expander functions rely on the mathematics developed for the Hessian-definiteness invariance theorem for linear projection transformations of variables, described in authors' paper, to generate the full, high-dimensional gradient and Hessian from the lower-dimensional derivative objects. This greatly relieves the computational burden of generating the regression-function derivatives, which in turn can be fed into any optimization routine that utilizes such derivatives. The theorem guarantees that Hessian definiteness is preserved, meaning that reasoning about this property can be performed in the low-dimensional space of the base distribution. This is often a much easier task than its equivalent in the full, high-dimensional space. Definiteness of Hessian can be useful in selecting optimization/sampling algorithms such as Newton-Raphson optimization or its sampling equivalent, the Stochastic Newton Sampler. Finally, in addition to being a computational tool, the regression expansion framework is of conceptual value by offering new opportunities to generate novel regression problems.

Last updated 4 years ago

2.30 score 20 scripts 212 downloads

CFC - Cause-Specific Framework for Competing-Risk Analysis

Numerical integration of cause-specific survival curves to arrive at cause-specific cumulative incidence functions, with three usage modes: 1) Convenient API for parametric survival regression followed by competing-risk analysis, 2) API for CFC, accepting user-specified survival functions in R, and 3) Same as 2, but accepting survival functions in C++. For mathematical details and software tutorial, see Mahani and Sharabiani (2019) <DOI:10.18637/jss.v089.i09>.

Last updated 2 years ago

cppopenmp

2.11 score 13 scripts 266 downloads

DBR - Discrete Beta Regression

Bayesian Beta Regression, adapted for bounded discrete responses, commonly seen in survey responses. Estimation is done via Markov Chain Monte Carlo sampling, using a Gibbs wrapper around univariate slice sampler (Neal (2003) <DOI:10.1214/aos/1056562461>), as implemented in the R package MfUSampler (Mahani and Sharabiani (2017) <DOI: 10.18637/jss.v078.c01>).

Last updated 2 years ago

2.04 score 11 scripts 234 downloads

MatchLinReg - Combining Matching and Linear Regression for Causal Inference

Core functions as well as diagnostic and calibration tools for combining matching and linear regression for causal inference in observational studies.

Last updated 2 years ago

2.00 score 6 scripts 175 downloads

EnsemblePCReg - Extensible Package for Principal-Component-Regression-Based Heterogeneous Ensemble Meta-Learning

Extends the base classes and methods of 'EnsembleBase' package for Principal-Components-Regression-based (PCR) integration of base learners. Default implementation uses cross-validation error to choose the optimal number of PC components for the final predictor. The package takes advantage of the file method provided in 'EnsembleBase' package for writing estimation objects to disk in order to circumvent RAM bottleneck. Special save and load methods are provided to allow estimation objects to be saved to permanent files on disk, and to be loaded again into temporary files in a later R session. Users and developers can extend the package by extending the generic methods and classes provided in 'EnsembleBase' package as well as this package.

Last updated 3 years ago

openjdk

2.00 score 216 downloads

EnsembleBase - Extensible Package for Parallel, Batch Training of Base Learners for Ensemble Modeling

Extensible S4 classes and methods for batch training of regression and classification algorithms such as Random Forest, Gradient Boosting Machine, Neural Network, Support Vector Machines, K-Nearest Neighbors, Penalized Regression (L1/L2), and Bayesian Additive Regression Trees. These algorithms constitute a set of 'base learners', which can subsequently be combined together to form ensemble predictions. This package provides cross-validation wrappers to allow for downstream application of ensemble integration techniques, including best-error selection. All base learner estimation objects are retained, allowing for repeated prediction calls without the need for re-training. For large problems, an option is provided to save estimation objects to disk, along with prediction methods that utilize these objects. This allows users to train and predict with large ensembles of base learners without being constrained by system RAM.

Last updated 8 years ago

openjdk

1.95 score 3 dependents 5 scripts 213 downloads

BSGW - Bayesian Survival Model with Lasso Shrinkage Using Generalized Weibull Regression

Bayesian survival model using Weibull regression on both scale and shape parameters. Dependence of shape parameter on covariates permits deviation from proportional-hazard assumption, leading to dynamic - i.e. non-constant with time - hazard ratios between subjects. Bayesian Lasso shrinkage in the form of two Laplace priors - one for scale and one for shape coefficients - allows for many covariates to be included. Cross-validation helper functions can be used to tune the shrinkage parameters. Monte Carlo Markov Chain (MCMC) sampling using a Gibbs wrapper around Radford Neal's univariate slice sampler (R package MfUSampler) is used for coefficient estimation.

Last updated 2 years ago

1.00 score 1 stars 8 scripts 235 downloads

SAMUR - Stochastic Augmentation of Matched Data Using Restriction Methods

Augmenting a matched data set by generating multiple stochastic, matched samples from the data using a multi-dimensional histogram constructed from dropping the input matched data into a multi-dimensional grid built on the full data set. The resulting stochastic, matched sets will likely provide a collectively higher coverage of the full data set compared to the single matched set. Each stochastic match is without duplication, thus allowing downstream validation techniques such as cross-validation to be applied to each set without concern for overfitting.

Last updated 2 years ago

1.00 score 3 scripts 98 downloads

EnsemblePenReg - Extensible Classes and Methods for Penalized-Regression-Based Integration of Base Learners

Extending the base classes and methods of EnsembleBase package for Penalized-Regression-based (Ridge and Lasso) integration of base learners. Default implementation uses cross-validation error to choose the optimal lambda (shrinkage parameter) for the final predictor. The package takes advantage of the file method provided in EnsembleBase package for writing estimation objects to disk in order to circumvent RAM bottleneck. Special save and load methods are provided to allow estimation objects to be saved to permanent files on disk, and to be loaded again into temporary files in a later R session. Users and developers can extend the package by extending the generic methods and classes provided in EnsembleBase package as well as this package.

Last updated 8 years ago

openjdk

1.00 score 144 downloads

EnsembleCV - Extensible Package for Cross-Validation-Based Integration of Base Learners

Extends the base classes and methods of EnsembleBase package for cross-validation-based integration of base learners. Default implementation calculates average of repeated CV errors, and selects the base learner / configuration with minimum average error. The package takes advantage of the file method provided in EnsembleBase package for writing estimation objects to disk in order to circumvent RAM bottleneck. Special save and load methods are provided to allow estimation objects to be saved to permanent files on disk, and to be loaded again into temporary files in a later R session. The package can be extended, e.g. by adding variants of the current implementation.

Last updated 8 years ago

openjdk

1.00 score 1 stars 145 downloads

BayesMixSurv - Bayesian Mixture Survival Models using Additive Mixture-of-Weibull Hazards, with Lasso Shrinkage and Stratification

Bayesian Mixture Survival Models using Additive Mixture-of-Weibull Hazards, with Lasso Shrinkage and Stratification. As a Bayesian dynamic survival model, it relaxes the proportional-hazard assumption. Lasso shrinkage controls overfitting, given the increase in the number of free parameters in the model due to presence of two Weibull components in the hazard function.

Last updated 8 years ago

1.00 score 5 scripts 204 downloads