Initial commit

.Rbuildignore

@@ -0,0 +1,2 @@
+^.*\.Rproj$
+^\.Rproj\.user$

.Rproj.user/D4EE1C3C/cpp-definition-cache

@@ -0,0 +1,2 @@
+[
+]

.Rproj.user/D4EE1C3C/pcs/debug-breakpoints.pper

@@ -0,0 +1,6 @@
+{
+    "debugBreakpointsState" : {
+        "breakpoints" : [
+        ]
+    }
+}

.Rproj.user/D4EE1C3C/pcs/files-pane.pper

@@ -0,0 +1,9 @@
+{
+    "path" : "~/Dropbox/continuous GPS/GPSmatching/R",
+    "sortOrder" : [
+        {
+            "ascending" : true,
+            "columnIndex" : 2
+        }
+    ]
+}

.Rproj.user/D4EE1C3C/pcs/source-pane.pper

@@ -0,0 +1,3 @@
+{
+    "activeTab" : 2
+}

.Rproj.user/D4EE1C3C/pcs/windowlayoutstate.pper

@@ -0,0 +1,14 @@
+{
+    "left" : {
+        "panelheight" : 862,
+        "splitterpos" : 361,
+        "topwindowstate" : "NORMAL",
+        "windowheight" : 903
+    },
+    "right" : {
+        "panelheight" : 862,
+        "splitterpos" : 541,
+        "topwindowstate" : "NORMAL",
+        "windowheight" : 903
+    }
+}

.Rproj.user/D4EE1C3C/pcs/workbench-pane.pper

@@ -0,0 +1,6 @@
+{
+    "TabSet1" : 3,
+    "TabSet2" : 3,
+    "TabZoom" : {
+    }
+}

.Rproj.user/D4EE1C3C/persistent-state

@@ -0,0 +1,8 @@
+build-last-errors="[]"
+build-last-errors-base-dir="~/Dropbox/continuous GPS/GPSmatching/"
+build-last-outputs="[{\"output\":\"==> R CMD INSTALL --preclean --no-multiarch --with-keep.source GPSmatching\\n\\n\",\"type\":0},{\"output\":\"* installing to library ‘/Library/Frameworks/R.framework/Versions/3.5/Resources/library’\\n\",\"type\":1},{\"output\":\"\",\"type\":1},{\"output\":\"* installing *source* package ‘GPSmatching’ ...\\n\",\"type\":1},{\"output\":\"\",\"type\":1},{\"output\":\"** R\\n\",\"type\":1},{\"output\":\"** byte-compile and prepare package for lazy loading\\n\",\"type\":1},{\"output\":\"\",\"type\":1},{\"output\":\"** help\\n\",\"type\":1},{\"output\":\"\",\"type\":1},{\"output\":\"*** installing help indices\\n\",\"type\":1},{\"output\":\"** building package indices\\n\",\"type\":1},{\"output\":\"\",\"type\":1},{\"output\":\"** testing if installed package can be loaded\\n\",\"type\":1},{\"output\":\"\",\"type\":1},{\"output\":\"* DONE (GPSmatching)\\n\",\"type\":1},{\"output\":\"\",\"type\":1}]"
+compile_pdf_state="{\"errors\":[],\"output\":\"\",\"running\":false,\"tab_visible\":false,\"target_file\":\"\"}"
+files.monitored-path=""
+find-in-files-state="{\"handle\":\"\",\"input\":\"\",\"path\":\"\",\"regex\":true,\"results\":{\"file\":[],\"line\":[],\"lineValue\":[],\"matchOff\":[],\"matchOn\":[]},\"running\":false}"
+imageDirtyState="1"
+saveActionState="-1"

.Rproj.user/D4EE1C3C/rmd-outputs

@@ -0,0 +1,5 @@
+
+
+
+
+

.Rproj.user/D4EE1C3C/saved_source_markers

@@ -0,0 +1 @@
+{"active_set":"","sets":[]}

.Rproj.user/D4EE1C3C/sources/prop/10859896

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "7,1",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/22C1E8DB

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "6,16",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/2603739E

@@ -0,0 +1,2 @@
+{
+}

.Rproj.user/D4EE1C3C/sources/prop/2A3EF788

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "16,0",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/2B5B49F0

@@ -0,0 +1,2 @@
+{
+}

.Rproj.user/D4EE1C3C/sources/prop/337CD24F

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "26,51",
+    "scrollLine" : "9"
+}

.Rproj.user/D4EE1C3C/sources/prop/3A8D2CFC

@@ -0,0 +1,2 @@
+{
+}

.Rproj.user/D4EE1C3C/sources/prop/3C0F604E

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "5,0",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/3E36F6D7

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "3,25",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/44BB7997

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "16,4",
+    "scrollLine" : "5"
+}

.Rproj.user/D4EE1C3C/sources/prop/6547FC1E

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "14,17",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/81C40F4B

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "4,0",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/93AC7835

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "36,91",
+    "scrollLine" : "29"
+}

.Rproj.user/D4EE1C3C/sources/prop/95C028E1

@@ -0,0 +1,5 @@
+{
+    "cursorPosition" : "16,22",
+    "scrollLine" : "4",
+    "tempName" : "Untitled1"
+}

.Rproj.user/D4EE1C3C/sources/prop/9C407BEA

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "10,32",
+    "scrollLine" : "5"
+}

.Rproj.user/D4EE1C3C/sources/prop/B3B4CE9F

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "30,1",
+    "scrollLine" : "16"
+}

.Rproj.user/D4EE1C3C/sources/prop/C0F9D2EC

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "0,77",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/C315FAD

@@ -0,0 +1,5 @@
+{
+    "cursorPosition" : "13,12",
+    "scrollLine" : "0",
+    "tempName" : "Untitled1"
+}

.Rproj.user/D4EE1C3C/sources/prop/C3A20D2A

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "9,13",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/D21B565D

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "8,0",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/D48D6DC5

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "11,10",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/D760D727

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "48,0",
+    "scrollLine" : "24"
+}

.Rproj.user/D4EE1C3C/sources/prop/F0B69EF1

@@ -0,0 +1,4 @@
+{
+    "cursorPosition" : "4,12",
+    "scrollLine" : "0"
+}

.Rproj.user/D4EE1C3C/sources/prop/INDEX

@@ -0,0 +1,23 @@
+~%2FDesktop%2FGPSmatching-package.Rd="2B5B49F0"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FDESCRIPTION="D21B565D"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FNAMESPACE="81C40F4B"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FR%2FCovariate_balance.R="C0F9D2EC"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FR%2FGPSmatching-package.R="6547FC1E"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FR%2FGPSmatching.R="D48D6DC5"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FR%2FGPSmatching_smooth.R="3E36F6D7"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2FR%2FoptimIA-package.R="3A8D2CFC"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2FGPSmatching-package.Rd="C315FAD"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2FGPSmatching.Rd="95C028E1"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2FGPSmatching_smooth.Rd="9C407BEA"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2Fabsolute_corr_fun.Rd="2A3EF788"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2Fcreate_matching.Rd="F0B69EF1"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2Fmatching_l1.Rd="B3B4CE9F"
+~%2FDropbox%2Fcontinuous%20GPS%2FGPSmatching%2Fman%2Fmatching_smooth.Rd="10859896"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FApplication%2FCovariate_balance.R="93AC7835"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FCovariate_balance.R="44BB7997"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FGPSmatching.R="22C1E8DB"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FOdessey%2FBenchmark200.R="337CD24F"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FOdessey%2Fbalance1000.R="3C0F604E"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FOdessey%2Fbalance200.R="C3A20D2A"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2FTest_balance.R="2603739E"
+~%2FDropbox%2Fcontinuous%20GPS%2FResubmission%2Fdata_generate.R="D760D727"

.Rproj.user/D4EE1C3C/sources/s-2EABFB91/176BF097-contents

@@ -0,0 +1,54 @@
+library("wCorr")
+library("parallel")
+library(fst)
+library(data.table)
+library("xgboost")
+load("/nfs/home/X/xwu/shared_space/ci3_xwu/National_Causal/data2016/balance/covariates.RData")
+
+
+cor_origin<-c(abs(cor(prematch_data$treat,prematch_data$year,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$mean_bmi,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$smoke_rate,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$hispanic,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$pct_blk,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$medhouseholdincome,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$medianhousevalue,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$poverty,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$education,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$popdensity,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$pct_owner_occ,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$pct_female,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$pct_dual,method = c("spearman"))),
+              abs(cor(prematch_data$treat,prematch_data$mean_entry_age,method = c("spearman"))))
+
+cor_matched<-c(abs(cor(match_data$treat,match_data$year,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$mean_bmi,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$smoke_rate,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$hispanic,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$pct_blk,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$medhouseholdincome,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$medianhousevalue,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$poverty,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$education,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$popdensity,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$pct_owner_occ,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$pct_female,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$pct_dual,method = c("spearman"))),
+               abs(cor(match_data$treat,match_data$mean_entry_age,method = c("spearman"))))
+#match_data2 <- match_data_xgb[((nrow(match_data_xgb)/50)*(4-1)+1):((nrow(match_data_xgb)/50)*(30)),]
+order<-sort(as.numeric(cor_origin), index.return =T)$ix
+par(mar=c(7.1, 10.1, 2.1, 2.1))
+#par(mar=c(8.1,4.1,4.1,4.1))
+par(mgp=c(5,1,0))
+histcolors = c(rgb(0,0,1,.8), rgb(0,1,0,.8),rgb(1,0,0,.5),rgb(0.0,0.8,0,.5))
+plot((as.numeric(cor_origin))[order],c(1:length(order)),yaxt="n",xlab="Absolute Correlation",ylab="",type = "b",lty=2,pch=3,xlim=c(0,0.55), col = histcolors[1],main="",lwd=3,cex.lab=2.5,cex.axis=2,cex.main=3)
+#lines((as.numeric(cor_origin2))[order],c(1:length(order)), col = histcolors[1],lwd=3,lty=2,pch=1)
+lines((as.numeric(cor_matched))[order],c(1:length(order)), col = histcolors[2],lwd=3,lty=2,pch=1)
+#lines((as.numeric(cor_matched2))[order],c(1:length(order)), col = histcolors[2],lwd=3,pch=1)
+#lines(((cor_weight))[order],c(1:length(order)), col = histcolors[3],lty=2,lwd=3)
+#lines(((cor_weight2))[order],c(1:length(order)), col = histcolors[3],lwd=3)
+axis(2, at=c(1:length(order)), labels=c("year","mean_bmi","smoke_rate","hispanic","pct_blk","MdHsIcm","MdVIHs",
+                                        "poverty","education","popdensity","pct_owner_occ","pct_female","pct_dual","mean_entry_age")[order],cex.axis=1.5,las=1)
+legend("bottomright",legend =  c("original","matched (boost)"),
+       col = histcolors[c(1,2)], lty= c(1,2))
+abline(v=0.1,lty=3,lwd=3)

.Rproj.user/D4EE1C3C/sources/s-2EABFB91/1CF0D10F-contents

@@ -0,0 +1,24 @@
+#' The 'GPSmatching' package.
+#'
+#' @description An R package for implementing matching on generalized propensity scores with continuous exposures.
+#' @details We developed an innovative approach for estimating causal effects using observational data in settings with continuous exposures, and introduce a new framework for GPS caliper matching.
+#'
+#' @docType package
+#' @name GPSmatching-package
+#' @aliases GPSmatching
+#' @author Xiao Wu
+#' @import parallel
+#' @import data.table
+#' @import SuperLearner
+#' @import xgboost
+#' @import earth
+#' @import ranger
+#' @import gam
+#' @import KernSmooth
+#'
+#' @references
+#' Wu, X., Mealli, F., Kioumourtzoglou, M.A., Dominici, F. and Braun, D., 2018. Matching on generalized propensity scores with continuous exposures. arXiv preprint arXiv:1812.06575.
+#'
+#' Kennedy, E.H., Ma, Z., McHugh, M.D. and Small, D.S., 2017. Non‐parametric methods for doubly robust estimation of continuous treatment effects. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 79(4), pp.1229-1245.
+#'
+NULL

.Rproj.user/D4EE1C3C/sources/s-2EABFB91/1EBD073B-contents

@@ -0,0 +1,41 @@
+#' Estimate smoothed exposure-response function (ERF).
+#'
+#' @param matched_Y a vector of outcome variable in matched set.
+#' @param matched_w a vector of continuous exposure variable in matched set.
+#' @param bw.seq a vector of bandwidth values (Default is seq(0.2,2,0.2)).
+#' @param w.vals a vector of exposure levels that ERF curves was evaluated.
+#' @return
+#' \code{erf}: The function returns a vector saved the output values of exposure-response function (ERF) given input \code{w.vals}.
+#' @export
+
+# Fit non-parametric kernel smoothing on matched set
+matching_smooth<-function(matched_Y,
+                          matched_w,
+                          bw.seq=seq(0.2,2,0.2),
+                          w.vals){
+  ## The specified Gaussian kernel
+  kern_fun <- function(t){ dnorm(t) }
+  w_fun <- function(bw){
+    w.avals <- NULL
+    for (w.val in w.vals){
+    w.std <- (matched_w-w.val)/bw
+    kern.std <- kern_fun(w.std)/bw
+    w.avals <- c(w.avals, mean(w.std^2*kern.std)*(kern(0)/bw) /
+                   (mean(kern.std)*mean(w.std^2*kern.std)-mean(w.std*kern.std)^2))
+      }
+    return(w.avals/length(matched_w))
+    }
+  hatvals <- function(bw){approx(w.vals,w_fun(bw),xout=matched_w,rule=2)$y}
+  smooth_fun <- function(out,bw){
+    approx(locpoly(w,out,bandwidth=bw, gridsize=1000),xout=matched_w,rule=2)$y
+    }
+  ##
+  risk_fun <- function(h){
+    hats <- hatvals(h); mean( ((matched_Y - smooth_fun(matched_Y,bw=h))/(1-hats))^2)
+    }
+  risk.val <- sapply(bw.seq, risk_fun)
+  h.opt <- bw.seq[which.min(risk.val)]
+
+  erf <- approx(locpoly(matched_w, matched_Y, bandwidth=h.opt), xout=w.vals)$y
+  return(erf)
+}

.Rproj.user/D4EE1C3C/sources/s-2EABFB91/1FD38C45-contents

@@ -0,0 +1,41 @@
+#' Estimate smoothed exposure-response function (ERF).
+#'
+#' @param matched_Y a vector of outcome variable in matched set.
+#' @param matched_w a vector of continuous exposure variable in matched set.
+#' @param bw.seq a vector of bandwidth values (Default is seq(0.2,2,0.2)).
+#' @param w.vals a vector of exposure levels that ERF curves was evaluated.
+#' @return
+#' \code{erf}: The function returns a vector saved the output values of exposure-response function (ERF) given input \code{w.vals}.
+#' @export
+
+# Fit non-parametric kernel smoothing on matched set
+matching_smooth<-function(matched_Y,
+                          matched_w,
+                          bw.seq=seq(0.2,2,0.2),
+                          w.vals){
+  ## The specified Gaussian kernel
+  kern_fun <- function(t){ dnorm(t) }
+  w_fun <- function(bw){
+    w.avals <- NULL
+    for (w.val in w.vals){
+    w.std <- (matched_w-w.val)/bw
+    kern.std <- kern_fun(w.std)/bw
+    w.avals <- c(w.avals, mean(w.std^2*kern.std)*(kern_fun(0)/bw) /
+                   (mean(kern.std)*mean(w.std^2*kern.std)-mean(w.std*kern.std)^2))
+      }
+    return(w.avals/length(matched_w))
+    }
+  hatvals <- function(bw){approx(w.vals,w_fun(bw),xout=matched_w,rule=2)$y}
+  smooth_fun <- function(out,bw){
+    approx(locpoly(matched_w,out,bandwidth=bw, gridsize=1000),xout=matched_w,rule=2)$y
+    }
+  ##
+  risk_fun <- function(h){
+    hats <- hatvals(h); mean( ((matched_Y - smooth_fun(matched_Y,bw=h))/(1-hats))^2)
+    }
+  risk.val <- sapply(bw.seq, risk_fun)
+  h.opt <- bw.seq[which.min(risk.val)]
+
+  erf <- approx(locpoly(matched_w, matched_Y, bandwidth=h.opt), xout=w.vals)$y
+  return(erf)
+}