Skip to content

Latest commit

 

History

History
312 lines (217 loc) · 9.84 KB

NEWS.md

File metadata and controls

312 lines (217 loc) · 9.84 KB
Raw

CausalGPS 0.5.0 (2024-06-19)

Added

  • estimate_gps now accepts formula for generating GPS object.
  • estimate_gps now has functions for generic plot, print, and summary functions.
  • trim_it function.

Removed

  • preprocess_data function.
  • trim_gps function.

CausalGPS 0.4.2 (2024-04-13)

Added

  • heads-up message about changes in the software design that will be implemented in CausalGPS version 0.5.x.

CausalGPS 0.4.1 (2023-09-29)

Added

  • Extra step to check consistency of delta_n with exposure range.
  • Software paper examples were added.

Changed

  • Plotting pseudo population includes object details. Set include_details = TRUE.
    *generate_pseudo_pop does not take Y as an input.

CausalGPS 0.4.0 (2023-05-25)

Changed

  • Docker image supports R 4.2.3
  • generate_syn_data supports vectorized_y to accelerate data generation.
  • matching_fun --> dist_measure
  • matching_l1 --> matching_fn
  • estimate_semipmetric_erf now takes the gam models optional arguments.
  • estimate_pmetric_erf now takes the gnm models optional arguments.
  • trim_quantiles --> exposure_trim_qtls
  • generate_pseudo_pop function accepts gps_obj as an optional input.
  • internal_use is not part of parameters for estimate_gps function.
  • estimate_gps function only returns id, w, and computed gps as part of dataset.
  • Now the design and analysis phases are explicitly separated.
  • gps_model --> gps_density. Now it takes, normal and kernel options instead of parametric and non-parametric options.

Added

  • estimate_npmetric_erf supports both locpol and KernSmooth approaches.
  • There is gps_trim_qtls input parameter to trim data samples based on gps values.
  • Now users can also collect the original data in the pseudo population object.

Fixed

  • A bug with swapping transformed covairates with original one.

CausalGPS 0.3.1 (2023-05-15)

Changed

  • Some of unit tests have less accuracy to overcome the bug with stats::density function.

CausalGPS 0.3.0 (2023-02-15)

Changed

  • Unit tests support new wCorr release (#193).
  • Only optimized compilation is supported. In the previous versions, this approach is known as optimzied_compile == TRUE.

Added

  • The earth package is part of suggested packages.

CausalGPS 0.2.9 (2022-12-16)

Fixed

  • fixed a bug based on covariate balance threshold (#178, @naeemkh).
  • estimate_npmetric_erf assigns user-defined log file.

Changed

  • The process now prints the progress message based on the selected thresholds.
  • In estimate_npmetric_erf:
    • matched_Y --> m_Y
    • matched_w --> m_w
    • matched_cw --> counter_weight
  • In estimate_npmetric_erf function, the matched_cw input is now mandatory.
  • Internal kernel smoothing now uses locpol::locpol function.
  • The entire data set is trimmed based on trimming quantiles.
  • earth and ranger are not installed automatically. They can be installed manually if needed.
  • sysdata.rda is modified to reflect transition from counter and ipw to counter_weight
  • counter_weight is used as a counter or weight, in matching or weighting approaches. counter and ipw are dropped.
  • sl_lib becomes a required argument.
  • The package has been transferred into NSAPH-Software Github account.
  • Summary function of gpsm_pspop S3 object returns details of the adjusting process.

Added

  • Now Kolmogorov-Smirnov(KS) statistics are provided for the computed pseudo population.
  • effect size for the generated pseudo population is computed and reported.
  • Binary search approach is used when scale = 1.
  • pseodo_pop also includes covariate column names.
  • compute_closest_wgps_helper_no_sc is added to take care of the mostly used special case (scale = 1).

Removed

  • Dropped importing KernSmooth and tidyr packages.
  • pred_model argument dropped. The package only predicts using SuperLearner.

CausalGPS 0.2.8 (2022-06-22)

Fixed

  • Message for not implemented methods changed to reduce misunderstanding.
  • Empty counter will raise error in estimating non-parametric response function.

Changed

  • matching_l1 returns frequency table instead of entire vector.
  • Vectorized population compilation and used data.table for multi-thread assignment.
  • Removed nested parallelism in compiling pseudo population, which results in close control on memory.
  • estimate_npmetric_erf also returns optimal h and risk values.

Added

  • estimate_gps returns the optimal hyperparameters.
  • estimate_gps returns S3 object.
  • Internal xgboost approach support verbose parameter.
  • Pseudo-population object now report the parameters that are used for the best covariate balance.

CausalGPS 0.2.7 (2022-02-04)

Fixed

  • Naming covariate balance scores.

Changed

  • Restarting adaptive approach to keep trying up to maximum attempt.

Added

  • Synthetic data (synthetic_us_2010)
  • Check on not defined covariate balance (absolute_corr_fun, absolute_weighted_corr_fun)
  • Covariate balance threshold type: mean, median, maximal.
  • Improved test coverage.
  • Singularity definition file.

CausalGPS 0.2.6 (2021-09-06)

Added

  • added the status of optimized compile to generate_pseudo_pop function output.
  • compute_closest_wgps accepts the number of user-defined threads.

Changed

  • Vignette file names.
  • The trim condition from > and < into >= and <=.
  • Removed seed input from generate_syn_data function. In R package, setting seed value inside function is not recommended. Users can set the seed before using the function.
  • OpenMP uses user defined number of cores.

Fixed

  • Initial covariate balance for weighted approach. The counter column was not preallocated correctly.
  • Counter value for compiling. The initial value was set to one, which, however, zero is the correct one.
  • Private variable issue with OpenMP.
  • Fixed OpenMP option on macOS checks.

Removed

CausalGPS 0.2.5 (2021-07-23)

Changed

  • User needs to activate the logger

Fixed

  • CRAN package URLs are in canonical forms.

CausalGPS 0.2.4 (2021-07-11)

Added

  • OpenMP for Rcpp code
  • optimized_compile
  • log_system_info()
  • Frequently asked questions
  • logo

Changed

  • estimate_gps.Rmd
  • estimate_semi_erf -> estimate_semipmetric_erf
  • estimate_erf -> estimate_npmetric_erf
  • estimate_hr -> estimate_pmetric_erf
  • gen_pseudo_pop -> generate_pseudo_pop
  • gen_syn_data -> generate_syn_data
  • estimate_erf accepts counter as an input
  • estimate_erf can use multiple cores
  • generating_pseudo_population.Rmd
  • estimate_erf function description
  • estimate_hr function description
  • estimate_semi_erf function description
  • compute_risk function description and return value
  • outcome_models.Rmd
  • generate_synthetic_data.Rmd

Fixed

  • Rcpp parLapply worker processors arguments

Removed

  • running_appr

CausalGPS 0.2.3 (2021-05-12)

Fixed

  • Fixed documentations

CausalGPS 0.2.2 (2021-05-12)

Added

  • estimate_semi_erf
  • estimate_hr

Changed

  • Package name: GPSmatching --> CausalGPS

GPSmatching 0.2.1 (2021-04-23)

Added

  • User defined bin sequence in compiling pseudo population.
  • Non-parametric option for estimating GPS.
  • Adaptive approach to transform features in training sessions.
  • Cpp code for computing pair of w and GPS.
  • set_logger function.
  • Customized wrapper for ranger package.
  • Extended plot function for gen_pseudo_pop object (plot.R).
  • Extended plot function for estimate_erf object (plot.R).
  • Extended print function for estimate_erf object (print.R).
  • test-estimate_erf.R.
  • create_weighting.R.
  • Steps for adding test data into 'sysdata.rda'.
  • weighting option as causal inference approach.
  • absolute_weighted_corr_fun.R
  • Testing and running example guidelines for developers
  • Customized wrapper for xgboost package.
  • param as an argument to accept hyperparameters from users.

Changed

  • R dependency 2.7 --> 3.5
  • mclapply --> parLapply
  • estimate_erf output returns S3 object.
  • test-Covariate_balance.R --> test-absolute_corr_fun.R
  • covariate_balance.R --> absolute_corr_fun.R
  • User needs to pass m_xgboost instead of SL.xgboost to use XGBoost package for prediction purposes.

Fixed

  • mclapply memory issue (compute_closest_wgps.R).

GPSmatching 0.2.0 (2021-03-01)

Added

  • Covariate balance check for categorical data.
  • Contribution guidelines
  • Parallel flag in training models (mcSuperLearner)
  • gen_syn_data function for generating synthetic data
  • Unittest for gen_syn_data
  • Function to compute residuals and unittest
  • Function to impute NA values based on density and unittest
  • Function to separate prediction model training (train_it)
  • Function to separate min and max value estimation and unittest
  • Function to find the closest data based on GPS and w
  • Wrapper function to generate pseudo population and test it for covariate balance (gen_pseudo_pop)
  • Function to estimate only GPS value (estimate_gps)
  • Helper function to take the input data + GPS values and return pseudo population based on selected causal inference approach. The output of this function may or may not satisfy the covariate balance test. (compile_pseudo_pop)
  • check_args function to check availability of the required parameters.
  • check_covar_balance function to check if the generated pseudo population statistically acceptable.
  • create_matching function to generate pseudo population based on matching approach.
  • acknowledgments to index file

Changed

  • create_matching only generates matched dataset.
  • Covariate_balance.R --> covariate_balance.R
  • matching_smooth --> estimate_erf.R
  • risk_fun --> compute_risk
  • smooth_fun --> smooth_erf
  • hatvals --> estimate_hat_vals
  • kernel_fun --> generate_kernel
  • GPSmatching-package.R --> gpsmatching_package.R
  • GPSmatching_smooth.R --> gpsmatching_smooth.R

Removed

  • GPSmatching.R functions are separated into smaller functions, and the file is removed.