Graeme Blair

[WIP] test of wallet functionality

5

This PR is a proof-of-concept of "wallet" functionality - steps that add objects to an environment. The main use case for now is declaring a statistical model. This PR enables...

capture estimation errors

3

For lots of models there are errors in a small proportion (or even a large proportion) of simulations, i.e. many MLE and MCMC procedures. I think we don't want to...

Issue loading design from rds

2

Here we make a design, save it, restart R, then load it again. It doesn't work if we load tidyverse before Declaredesign, But it does work if we load DeclareDesign...

bug in subset and weights for lm handler (due to recent NSE changes)?

1

``` > declare_population(N = 100, x = rnorm(N)) + declare_estimator(x ~ 1, model = lm, subset = x > 1) Error in ~x > 1 : object 'x' not found...

library(future) issues in RStudio

14

There's seems to be some kind of bad interaction between `library(future)` and R 3.6.0 and our use of `future_lapply()`. @nfultz or others can you spot anything wrong here? The following...

many-to-many warning triggered incorrectly(?)

1

``` design

so that `map(designs, draw_data) %>% bind_rows(.id = "design")` will be appropriately labeled

more naming problems in expand_design

5

``` my_designer

rma problem

3

Meta analysis with the metafor package does not (and I think has not) worked with our declare_estimator infrastructure. It seems to be the case that when you do the following,...