GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.2.4

• Changed the maintainer to @boyuren158
• Updated arXiv link in all files to be compatible with CRAN style

Scientific Software - Peer-reviewed - R
Published by boyuren158 almost 2 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.2.3

• estimate_cerf_nngp takes outcome_col, treatment_col, and covariates_col names as inputs.
• estimate_cerf_gp takes outcome_col, treatment_col, and covariates_col names as inputs.
• estimate_cerf_gp and estimate_cerf_nngp have notes on selecting w.

Scientific Software - Peer-reviewed - R
Published by Naeemkh almost 2 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.2.2

• Addressed CRAN comment regarding fixing a typo in documentation.

Scientific Software - Peer-reviewed - R
Published by Naeemkh about 2 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.2.1

• Updated the C++ default version to resolve CRAN notes (from C++11 to C++17).
• Enhanced internal logging with more detailed messages.
• Incorporated the wCorr package for calculating weighted covariance balance.
• Renamed the 'traingps' function to 'estimategps'.
• Improved the summary function for S3 objects.

Scientific Software - Peer-reviewed - R
Published by Naeemkh almost 3 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.2.0

• estimatenoisenn now allows for parallelization with an added argument nthread for the number of CPUs used in parallel.
• estimatemeansd_nn now only computes the posterior variance.
• findoptimalnn now returns the posterior mean and covariate balance for the optimal hyper-parameter values.
• Add an argument kernel_fn to all nn related functions to allow for user-defined kernel functions.
• Add an argument formula to all nn related functions to allow for user-defined design matrix.
• findoptimalnn becomes an internal function.
• estimatenoisegp and estimatenoisenn become internal functions.
• estimatemeansd_nn becomes an internal function.
• computeweightgp becomes an internal function.
• computewcorr accepts w and confounders separately. It also normalizes w internally.
  
• computeposteriorsd_nn becomes an internal function.
• computeposteriorm_nn becomes an internal function.
• computederiveweights_gp becomes an internal function.
• computemsigma becomes an internal function.
• compute_inverse becomes an internal function.
• In computemsigma, tuning option does not have a default value.
• train_gps does not have default values.
• train_gps accepts vector of the SuperLearner package's libraries.
• trainGPS -> traingps

Scientific Software - Peer-reviewed - R
Published by Naeemkh about 3 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.1.0

First CRAN release.

Scientific Software - Peer-reviewed - R
Published by Naeemkh over 3 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - v0.0.2

All functions added to the package. However, have not been optimized.

Scientific Software - Peer-reviewed - R
Published by Naeemkh almost 4 years ago

GPCERF - An R package for implementing Gaussian processes for estimating causal exposure response curves - Pre-alpha v 0.0.1

Refactored code of full gaussian process

Added the following functions with unittests and examples.

• esimtatecerfgp
• computemsigma
• compute_inverse
• computeweightgp

Scientific Software - Peer-reviewed - R
Published by Naeemkh almost 4 years ago