some edit

experiment_code/major_revision/rq3/rq3.r

@@ -56,9 +56,9 @@ cpd$test_inclusion = as.logical(cpd$test_inclusion)
 cpd$issue_tag = as.logical(cpd$issue_tag)
 
 
-cpd$repo_id = interaction("prj",cpd$repo_id)
+#cpd$repo_id = interaction("prj",cpd$repo_id)
 cpd$repo_id = as.factor(cpd$repo_id)
-cpd$pair = interaction("pair",cpd$pair)
+#cpd$pair = interaction("pair",cpd$pair)
 cpd$pair = as.factor(cpd$pair)
 cpd$final_check = as.factor(cpd$final_check)
 
@@ -90,75 +90,20 @@ biserial.cor(cpd$churn, cpd$test_inclusion, use = c("all.obs", "complete.obs"),
 library(pROC)
 # build Conditional logistic regression for paired data
 library("survival")
-library(mclogit)
-library(MuMIn)
-library(performance)
-library(DescTools)
-library(JWileymisc)
-library(rcompanion)
-library(pubh)
-library(descr)
-data("electors")
 
-
-#########################################################
-
-
-library(lme4)
-
-
-dd <- within(electors,party.time<-interaction(party,time))
-dd <- within(dd,class.time<-interaction(time,class))
 #########################################################
 # 1. submitter-level
 
-#-------------
 
 library(mlogit)
-dp.CLR1 <- clogit(prefer~ 
+dp.CLR1 <- mclogit(cbind(prefer,pair)~ 
                    log(prev_pullreqs+0.5)+ log(followers+0.5)
                  + first_pr_proj+ author_core + log(co_occur_strength+0.5)
-                 + strata(pair)
+                 + (1|repo_id)
+                ,
                  data=cpd
                  )
 
-
-dp.CLR1 <- glmer(prefer~ 
-                    log(prev_pullreqs+0.5)+ log(followers+0.5)
-                  + first_pr_proj+ author_core + log(co_occur_strength+0.5)
-                 + (1|repo_id),
-                  data=cpd,
-                 family="binomial"
-)
-
-model_performance(dp.CLR1)
-r2.mmclogit(dp.CLR1)
-
-logLik(dp.CLR1)
-logLik(dp.CLR0)
-
-as.vector(logLik(dp.CLR0) - logLik(dp.CLR1)) / logLik(dp.CLR0)
-
-#Pesudo R square
-#x = full model deviance
-#y = null model deviance
-#k = number of parameters
-
-x = logLik(dp.CLR1)
-y = logLik(dp.CLR0)
-k = 5
-
-PRsq<- function (x, y, k ) {
-  a= 1- (x/y)
-  b= 1- ((x-k)/y)
-  cat ("Pesudo R Square Estimates", "\n")
-  cat ("McFadden's:          ", a, "\n")
-  cat ("adjusted McFadden's: ", b, "\n")
-}
-
-PRsq(x,y,k)
-
-vif(dp.CLR1)
 summary(dp.CLR1)
 prob=predict(dp.CLR1, type='response')
 cpd$prob=prob
@@ -166,7 +111,16 @@ a = roc(prefer ~ prob, data = cpd)
 a
 
 
-prob=predict(dp.CLR1, type='risk')
+dp.CLR1 <- clogit(prefer~ 
+                     log(prev_pullreqs+0.5)+ log(followers+0.5)
+                   + first_pr_proj+ author_core + log(co_occur_strength+0.5)
+                  + strata(pair)
+                  +repo_id
+                   ,
+                   data=cpd
+)
+summary(dp.CLR1)
+prob=predict(dp.CLR, type='risk')
 cpd$prob=prob
 a = roc(prefer ~ prob, data = cpd)
 a
@@ -235,7 +189,7 @@ dp.CLR <- mclogit(cbind(prefer,pair) ~
                  + BT
                  + CLA
                  + CR
-                 + (1 | repo_id)
+                 #+ (1 | repo_id)
                  #+ final_check 
                  #+ log(total_checks+0.5)
                  #+ log(check_tools+0.5)