Add another robustness check for varying levels of accuracy.

[ml_measurement_error_public.git] / simulations / .Rhistory

diff --git a/simulations/.Rhistory b/simulations/.Rhistory
index 6de9aa0e9381b1093a99bc7c1df21cc8ba80989c..f0a360925f17f601956a2834be22250b3815ed71 100644 (file)
--- a/simulations/.Rhistory
+++ b/simulations/.Rhistory
@@ -63,3 +63,94 @@ list.files()
 install.packages("filelock")
 q()
 n
 install.packages("filelock")
 q()
 n

+df
+df
+outcome_formula <- y ~ x + z
+outcome_family=gaussian()
+proxy_formula <- w_pred ~ x
+truth_formula <- x ~ z
+params <- start
+ll.y.obs.x0
+ll.y.obs.x1
+rater_formula <- x.obs ~ x
+rater_formula
+rater.modle.matrix.obs.x0
+rater.model.matrix.obs.x0
+names(rater.model.matrix.obs.x0)
+head(rater.model.matrix.obs.x0)
+df.obs
+ll.x.obs.0
+rater.params
+rater.params %*% t(rater.model.matrix.x.obs.0[df.obs$xobs.0==1])
+df.obs$xobs.0==1
+df.obs$x.obs.0==1
+ll.x.obs.0[df.obs$x.obs.0==1]
+rater.model.matrix.x.obs.0[df.obs$x.obs.0==1,]
+df.obs$x.obs.0==1
+n.rater.model.covars <- dim(rater.model.matrix.x.obs.0)[2]
+        rater.params <- params[param.idx:n.rater.model.covars]
+rater.params
+        ll.x.obs.0[df.obs$x.obs.0==1] <- plogis(rater.params %*% t(rater.model.matrix.x.obs.0[df.obs$x.obs.0==1,]), log=TRUE)
+t(rater.model.matrix.x.obs.0[df.obs$x.obs.0==1,]
+)
+dimt(rater.model.matrix.x.obs.0[df.obs$x.obs.0==1,])
+dim(t(rater.model.matrix.x.obs.0[df.obs$x.obs.0==1,]))
+dim(ll.x.obs.0[df.obs$x.obs.0==1])
+rater.params
+rater.params
+rater.params
+rater_formula
+rater.params
+)
+1+1
+q()
+n
+outcome_formula <- y ~ x + z
+proxy_formula <- w_pred ~ x + z + y
+truth_formula <- x ~ z
+proxy_formula
+eyboardio Model 01 - Kaleidoscope locally built
+df <- df.triple.proxy.mle
+outcome_family='gaussian'
+outcome_family=gaussian()
+proxy_formulas=list(proxy_formula,x.obs.0~x, x.obs.1~x)
+proxy_formulas
+proxy_familites <- rep(binomial(link='logit'),3)
+proxy_families = rep(binomial(link='logit'),3)
+proxy_families
+proxy_families = list(binomial(link='logit'),binomial(link='logit'),binomial(link='logit'))
+proxy_families
+proxy_families[[1]]
+proxy.params
+i
+proxy_params
+proxy.params
+params
+params <- start
+df.triple.proxy.mle
+df
+coder.formulas <- c(x.obs.0 ~ x, x.obs.1 ~x)
+outcome.formula
+outcome_formula
+depvar(outcome_formula
+)
+outcome_formula$terms
+terms(outcome_formula)
+q()
+n
+df.triple.proxy.mle
+triple.proxy.mle
+df
+df <- df.triple.proxy
+outcome_family <- binomial(link='logit')
+outcome_formula <- y ~x+z
+proxy_formula <- w_pred ~ y
+coder_formulas=list(y.obs.1~y,y.obs.2~y); proxy_formula=w_pred~y; proxy_family=binomial(link='logit'))
+coder_formulas=list(y.obs.1~y,y.obs.2~y); proxy_formula=w_pred~y; proxy_family=binomial(link='logit')
+coder_formulas=list(y.obs.0~y,y.obs.1~y)
+traceback()
+df
+df
+outcome.model.matrix
+q()
+n