2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
wo_df_ranef |>
|
|
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= condval - (1.96 * condsd), ymax= condval + (1.96 * condsd))) +
|
|
|
|
|
theme_bw()
|
2024-05-07 23:40:38 +00:00
|
|
|
# this is the file with the lmer multi-level rddAnalysis
|
|
|
|
|
library(tidyverse)
|
|
|
|
|
library(plyr)
|
|
|
|
|
# 0 loading the readme data in
|
|
|
|
|
try(setwd(dirname(rstudioapi::getActiveDocumentContext()$path)))
|
|
|
|
|
readme_df <- read_csv("../final_data/deb_readme_did.csv")
|
|
|
|
|
# 1 preprocessing
|
|
|
|
|
#colnames(readme_df) <- c("upstream_vcs_link", "event_date", "event_hash", "before_all_ct", "before_mrg_ct", "after_all_ct", "after_mrg_ct", "before_auth_new", "after_commit_new", "after_auth_new", "before_commit_new")
|
|
|
|
|
col_order <- c("upstream_vcs_link", "age_of_project", "event_date", "event_hash", "before_all_ct", "after_all_ct", "before_mrg_ct", "after_mrg_ct", "before_auth_new", "after_auth_new", "before_commit_new", "after_commit_new")
|
|
|
|
|
readme_df <- readme_df[,col_order]
|
|
|
|
|
readme_df$ct_before_all <- str_split(gsub("[][]","", readme_df$before_all_ct), ", ")
|
|
|
|
|
readme_df$ct_after_all <- str_split(gsub("[][]","", readme_df$after_all_ct), ", ")
|
|
|
|
|
readme_df$ct_before_mrg <- str_split(gsub("[][]","", readme_df$before_mrg_ct), ", ")
|
|
|
|
|
readme_df$ct_after_mrg <- str_split(gsub("[][]","", readme_df$after_mrg_ct), ", ")
|
|
|
|
|
drop <- c("before_all_ct", "before_mrg_ct", "after_all_ct", "after_mrg_ct")
|
|
|
|
|
readme_df = readme_df[,!(names(readme_df) %in% drop)]
|
|
|
|
|
# 2 some expansion needs to happens for each project
|
|
|
|
|
expand_timeseries <- function(project_row) {
|
|
|
|
|
longer <- project_row |>
|
|
|
|
|
pivot_longer(cols = starts_with("ct"),
|
|
|
|
|
names_to = "window",
|
|
|
|
|
values_to = "count") |>
|
|
|
|
|
unnest(count)
|
|
|
|
|
longer$observation_type <- gsub("^.*_", "", longer$window)
|
|
|
|
|
longer <- ddply(longer, "observation_type", transform, week=seq(from=0, by=1, length.out=length(observation_type)))
|
|
|
|
|
longer$count <- as.numeric(longer$count)
|
|
|
|
|
#longer <- longer[which(longer$observation_type == "all"),]
|
|
|
|
|
return(longer)
|
2024-04-24 17:59:07 +00:00
|
|
|
}
|
2024-05-07 23:40:38 +00:00
|
|
|
expanded_data <- expand_timeseries(readme_df[1,])
|
|
|
|
|
for (i in 2:nrow(readme_df)){
|
|
|
|
|
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
|
2024-04-24 17:59:07 +00:00
|
|
|
}
|
2024-05-07 23:40:38 +00:00
|
|
|
#filter out the windows of time that we're looking at
|
|
|
|
|
window_num <- 8
|
|
|
|
|
windowed_data <- expanded_data |>
|
2024-04-24 17:59:07 +00:00
|
|
|
filter(week >= (27 - window_num) & week <= (27 + window_num)) |>
|
|
|
|
|
mutate(D = ifelse(week > 27, 1, 0))
|
2024-05-07 23:40:38 +00:00
|
|
|
#scale the age numbers
|
|
|
|
|
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
|
|
|
|
|
windowed_data$week_offset <- windowed_data$week - 27
|
|
|
|
|
#separate out the cleaning d
|
|
|
|
|
all_actions_data <- windowed_data[which(windowed_data$observation_type == "all"),]
|
|
|
|
|
mrg_actions_data <- windowed_data[which(windowed_data$observation_type == "mrg"),]
|
2024-05-09 22:05:21 +00:00
|
|
|
#find some EDA to identify which types of models might be the best for this
|
|
|
|
|
hist(log(all_actions_data$count))
|
2024-05-08 14:33:03 +00:00
|
|
|
all_actions_data$logged_count <- log(all_actions_data$count)
|
2024-05-07 23:40:38 +00:00
|
|
|
all_actions_data$log1p_count <- log1p(all_actions_data$count)
|
|
|
|
|
# 3 rdd in lmer analysis
|
|
|
|
|
# rdd: https://rpubs.com/phle/r_tutorial_regression_discontinuity_design
|
|
|
|
|
# lmer: https://www.youtube.com/watch?v=LzAwEKrn2Mc
|
|
|
|
|
library(lme4)
|
2024-04-23 14:55:37 +00:00
|
|
|
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
|
2024-05-07 23:40:38 +00:00
|
|
|
library(optimx)
|
|
|
|
|
library(lattice)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
2024-05-09 22:05:21 +00:00
|
|
|
summary(all_model)
|
2024-05-07 23:40:38 +00:00
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef <- ranef(all_model, condVar=TRUE)
|
|
|
|
|
dotplot(all_model_ranef)
|
2024-05-08 14:33:03 +00:00
|
|
|
df_ranefs <- as.data.frame(all_model_ranef)
|
|
|
|
|
D_df_ranef <- df_ranefs[which(df_ranefs$term == "D"),]
|
2024-05-09 22:05:21 +00:00
|
|
|
#below this groups the ranefs
|
2024-05-08 14:33:03 +00:00
|
|
|
has_zero <- function(condval, condsd){
|
|
|
|
|
bounds <- condsd * 1.96
|
|
|
|
|
return(ifelse(((condval - bounds) < 0),ifelse(((condval + bounds) > 0), 1, 0), 2))
|
|
|
|
|
}
|
|
|
|
|
df_ranefs <- df_ranefs |>
|
2024-05-09 22:05:21 +00:00
|
|
|
mutate(ranef_grouping = has_zero(condval, condsd)) |>
|
|
|
|
|
mutate(rank = rank(condval))
|
2024-05-08 14:33:03 +00:00
|
|
|
D_df_ranef <- df_ranefs[which(df_ranefs$term == "D"),]
|
|
|
|
|
hist(D_df_ranef$ranef_grouping)
|
2024-05-09 22:05:21 +00:00
|
|
|
D_df_ranef |>
|
|
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= condval - (1.96 * condsd), ymax= condval + (1.96 * condsd))) +
|
|
|
|
|
geom_bw()
|
2024-05-08 14:33:03 +00:00
|
|
|
#plot the ranefs
|
|
|
|
|
library(ggplot2)
|
|
|
|
|
D_df_ranef |>
|
2024-05-09 22:05:21 +00:00
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= condval - (1.96 * condsd), ymax= condval + (1.96 * condsd))) +
|
|
|
|
|
geom_bw()
|
2024-05-08 14:33:03 +00:00
|
|
|
D_df_ranef |>
|
|
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
2024-05-09 22:05:21 +00:00
|
|
|
geom_linerange(aes(ymin= condval - (1.96 * condsd), ymax= condval + (1.96 * condsd))) +
|
|
|
|
|
theme_bw()
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef <- ranef(all_model, condVar=TRUE)
|
|
|
|
|
dotplot(all_model_ranef)
|
|
|
|
|
df_ranefs <- as.data.frame(all_model_ranef)
|
|
|
|
|
#below this groups the ranefs
|
|
|
|
|
has_zero <- function(condval, condsd){
|
|
|
|
|
bounds <- condsd * 1.96
|
|
|
|
|
return(ifelse(((condval - bounds) < 0),ifelse(((condval + bounds) > 0), 1, 0), 2))
|
|
|
|
|
}
|
2024-05-08 14:33:03 +00:00
|
|
|
df_ranefs <- df_ranefs |>
|
|
|
|
|
mutate(ranef_grouping = has_zero(condval, condsd)) |>
|
|
|
|
|
mutate(rank = rank(condval))
|
|
|
|
|
D_df_ranef <- df_ranefs[which(df_ranefs$term == "D"),]
|
|
|
|
|
D_df_ranef |>
|
|
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
2024-05-09 22:05:21 +00:00
|
|
|
geom_linerange(aes(ymin= condval - (1.96 * condsd), ymax= condval + (1.96 * condsd))) +
|
|
|
|
|
theme_bw()
|
|
|
|
|
D_df_ranefs <- D_df_ranefs |>
|
|
|
|
|
mutate(rank = rank(condval))
|
|
|
|
|
D_df_ranef <- D_df_ranef |>
|
|
|
|
|
mutate(rank = rank(condval))
|
2024-05-08 14:33:03 +00:00
|
|
|
D_df_ranef |>
|
|
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
2024-05-09 22:05:21 +00:00
|
|
|
geom_linerange(aes(ymin= condval - (1.96 * condsd), ymax= condval + (1.96 * condsd))) +
|
|
|
|
|
theme_bw()
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_blup <- blup(all_model)
|
|
|
|
|
all_model_ranef <- ranef(all_model)
|
|
|
|
|
View(all_model_ranef)
|
|
|
|
|
df_ranefs <- as.data.frame(all_model_ranef)
|
|
|
|
|
dotplot(all_model_ranef)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_coef <- coef(all_model)
|
|
|
|
|
View(all_model_coef)
|
|
|
|
|
D_df_ranef <- df_ranefs[which(df_ranefs$term == "D"),]
|
|
|
|
|
D_df_ranef <- df_ranefs[which(df_ranefs$term == "D"),]
|
|
|
|
|
View(D_df_ranef)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
2024-04-25 01:55:56 +00:00
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
2024-05-09 22:05:21 +00:00
|
|
|
all_model_ranef <- ranef(all_model)
|
|
|
|
|
df_ranefs <- as.data.frame(all_model_ranef)
|
|
|
|
|
D_df_ranef <- df_ranefs[which(df_ranefs$term == "D"),]
|
|
|
|
|
View(D_df_ranef)
|
2024-05-07 23:40:38 +00:00
|
|
|
#identifying the quartiles of effect for D
|
2024-05-09 22:05:21 +00:00
|
|
|
all_model_variances <- postVar(all_model)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_variances <- vcov(all_model, condVar=TRUE)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
print(all_model_variances)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
conditional_variances_random <- lapply(all_model_variances, diag)
|
|
|
|
|
dotplot(conditional_variances_random)
|
|
|
|
|
dotplot(conditional_variances_random,
|
|
|
|
|
col = "blue",
|
|
|
|
|
pch = 19,
|
|
|
|
|
main = "Conditional Variances of Random Effects",
|
|
|
|
|
xlab = "Conditional Variance",
|
|
|
|
|
ylab = "Random Effect",
|
|
|
|
|
scales = list(x = list(log = TRUE)),
|
|
|
|
|
auto.key = list(space = "right"))
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_variances <- vcov(all_model, full=TRUE, condVar=TRUE)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
summary(all_model)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_variances <- vcov(all_model, full=TRUE, condVar=TRUE)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_variances <- varCorr(all_model)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_variances <- VarCorr(all_model)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
View(conditional_variances_random)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
attr(VarCorr(all_model)$upstream_vcs_link, "stddevs")^2
|
|
|
|
|
values <- attr(VarCorr(all_model)$upstream_vcs_link, "stddevs")^2
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_variances <- vcov(all_model)
|
|
|
|
|
View(all_model_variances)
|
|
|
|
|
print(all_model_variances)
|
|
|
|
|
all_model_ranef <- ranef(all_model)$upstream_vcs_link
|
|
|
|
|
View(all_model_ranef)
|
|
|
|
|
all_model_ranef <- cov(ranef(all_model))
|
|
|
|
|
random_effects <- ranef(all_model)
|
|
|
|
|
random_effects_variances <- lapply(random_effects$upstream_vcs_link, function(x) {
|
|
|
|
|
variances <- var(x$D:I(week_offset))
|
|
|
|
|
return(variances)
|
|
|
|
|
})
|
|
|
|
|
variances <- var(x$D)
|
|
|
|
|
summary_of_all <- summary(all_model)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
variance_components <- summary_of_all$varcor
|
|
|
|
|
View(variance_components)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
varcorr_of_all <- VarCorr(all_model)
|
|
|
|
|
View(varcorr_of_all)
|
|
|
|
|
print(varcorr_of_all)
|
|
|
|
|
all_coefficients <- coef(all_model)
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))
|
|
|
|
|
all_conf_intervals <- cbind(coefficients - 1.96 * standard_errors,
|
|
|
|
|
coefficients + 1.96 * standard_errors)
|
|
|
|
|
all_conf_intervals <- cbind(all_coefficients - 1.96 * all_standard_errors,
|
|
|
|
|
all_coefficients + 1.96 * all_standard_errors)
|
|
|
|
|
View(all_coefficients)
|
|
|
|
|
View(conditional_variances_random)
|
|
|
|
|
View(all_coefficients)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
confint(all_model)
|
|
|
|
|
all_coefficients <- coef(all_model)
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))[3]
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))[4]
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))[5]
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))[6]
|
|
|
|
|
all_standard_errors <- sqrt(diag(vcov(all_model)))[1]
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef <- ranef(all_model, condVar=TRUE)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
all_model_ranef <- ranef(all_model, condVar = FALSE)
|
|
|
|
|
View(all_model_ranef)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
dotplot(all_model_ranef)
|
|
|
|
|
dotplot(all_model_ranef_condvar)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
all_model_ranef_condvar[["upstream_vcs_link"]][["D"]]
|
|
|
|
|
View(all_model_ranef)
|
|
|
|
|
all_model_ranef_condvar$upstream_vcs_link
|
|
|
|
|
all_model_ranef_condvar$upstream_vcs_link$D
|
|
|
|
|
conditional_variances <- diag(vcov(model)$upstream_vcs_link$D)
|
|
|
|
|
conditional_variances <- diag(vcov(all_model)$upstream_vcs_link$D)
|
|
|
|
|
conditional_variances <- diag(vcov(all_model))
|
|
|
|
|
conditional_variances <- vcov(all_model)
|
|
|
|
|
View(conditional_variances)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- var(ranef(all_model, condVar = TRUE))
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- var(ranef(all_model, condVar = TRUE)$upstream_vcs_link$D)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)$upstream_vcs_link$D
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link$D, "condVar")
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "condVar")
|
|
|
|
|
df_ranefs <- as.data.frame(all_model_ranef_condvar)
|
|
|
|
|
View(df_ranefs)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
#all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
# optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "condVar")
|
|
|
|
|
#all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
# optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=TRUE)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "condVar")
|
|
|
|
|
df_ranefs <- as.data.frame(all_model_ranef_condvar)
|
|
|
|
|
View(df_ranefs)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
all_model_ranef <- ranef(all_model, condVar = FALSE)
|
|
|
|
|
View(all_model_ranef_condvar)
|
|
|
|
|
View(all_model_ranef_condvar[["upstream_vcs_link"]])
|
|
|
|
|
all_model_ranef_condvar[["upstream_vcs_link"]][["D"]]
|
|
|
|
|
View(all_model_ranef)
|
|
|
|
|
df_rn_no_cv <- as.data.frame(all_model_ranef)
|
|
|
|
|
View(df_rn_no_cv)
|
|
|
|
|
View(df_ranefs)
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link$D, "postVar")
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")[[4]]
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")[[3]]
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")[[2]]
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")[4]
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
isSingular(all_model)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (week_offset| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (week_offset| upstream_vcs_link), data=all_actions_data, REML=FALSE)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (I:(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D| upstream_vcs_link), data=all_actions_data, REML=FALSE)
|
|
|
|
|
summary_of_all <- summary(all_model)
|
|
|
|
|
summary(all_model)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
attr(all_model_ranef_condvar$upstream_vcs_link, "postVar")
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
|
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
# this is the file with the lmer multi-level rddAnalysis
|
2024-04-27 19:30:20 +00:00
|
|
|
library(tidyverse)
|
|
|
|
|
library(plyr)
|
2024-05-09 22:05:21 +00:00
|
|
|
# 0 loading the readme data in
|
2024-04-27 19:30:20 +00:00
|
|
|
try(setwd(dirname(rstudioapi::getActiveDocumentContext()$path)))
|
2024-05-09 22:05:21 +00:00
|
|
|
readme_df <- read_csv("../final_data/deb_readme_did.csv")
|
|
|
|
|
# 1 preprocessing
|
|
|
|
|
#colnames(readme_df) <- c("upstream_vcs_link", "event_date", "event_hash", "before_all_ct", "before_mrg_ct", "after_all_ct", "after_mrg_ct", "before_auth_new", "after_commit_new", "after_auth_new", "before_commit_new")
|
2024-04-27 19:30:20 +00:00
|
|
|
col_order <- c("upstream_vcs_link", "age_of_project", "event_date", "event_hash", "before_all_ct", "after_all_ct", "before_mrg_ct", "after_mrg_ct", "before_auth_new", "after_auth_new", "before_commit_new", "after_commit_new")
|
2024-05-09 22:05:21 +00:00
|
|
|
readme_df <- readme_df[,col_order]
|
|
|
|
|
readme_df$ct_before_all <- str_split(gsub("[][]","", readme_df$before_all_ct), ", ")
|
|
|
|
|
readme_df$ct_after_all <- str_split(gsub("[][]","", readme_df$after_all_ct), ", ")
|
|
|
|
|
readme_df$ct_before_mrg <- str_split(gsub("[][]","", readme_df$before_mrg_ct), ", ")
|
|
|
|
|
readme_df$ct_after_mrg <- str_split(gsub("[][]","", readme_df$after_mrg_ct), ", ")
|
2024-04-27 19:30:20 +00:00
|
|
|
drop <- c("before_all_ct", "before_mrg_ct", "after_all_ct", "after_mrg_ct")
|
2024-05-09 22:05:21 +00:00
|
|
|
readme_df = readme_df[,!(names(readme_df) %in% drop)]
|
2024-04-27 19:30:20 +00:00
|
|
|
# 2 some expansion needs to happens for each project
|
|
|
|
|
expand_timeseries <- function(project_row) {
|
|
|
|
|
longer <- project_row |>
|
|
|
|
|
pivot_longer(cols = starts_with("ct"),
|
|
|
|
|
names_to = "window",
|
|
|
|
|
values_to = "count") |>
|
|
|
|
|
unnest(count)
|
|
|
|
|
longer$observation_type <- gsub("^.*_", "", longer$window)
|
|
|
|
|
longer <- ddply(longer, "observation_type", transform, week=seq(from=0, by=1, length.out=length(observation_type)))
|
|
|
|
|
longer$count <- as.numeric(longer$count)
|
|
|
|
|
#longer <- longer[which(longer$observation_type == "all"),]
|
|
|
|
|
return(longer)
|
|
|
|
|
}
|
2024-05-09 22:05:21 +00:00
|
|
|
expanded_data <- expand_timeseries(readme_df[1,])
|
|
|
|
|
for (i in 2:nrow(readme_df)){
|
|
|
|
|
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
|
2024-04-27 19:30:20 +00:00
|
|
|
}
|
|
|
|
|
#filter out the windows of time that we're looking at
|
|
|
|
|
window_num <- 8
|
|
|
|
|
windowed_data <- expanded_data |>
|
|
|
|
|
filter(week >= (27 - window_num) & week <= (27 + window_num)) |>
|
|
|
|
|
mutate(D = ifelse(week > 27, 1, 0))
|
|
|
|
|
#scale the age numbers
|
|
|
|
|
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
|
|
|
|
|
windowed_data$week_offset <- windowed_data$week - 27
|
|
|
|
|
#separate out the cleaning d
|
|
|
|
|
all_actions_data <- windowed_data[which(windowed_data$observation_type == "all"),]
|
|
|
|
|
mrg_actions_data <- windowed_data[which(windowed_data$observation_type == "mrg"),]
|
2024-05-07 23:40:38 +00:00
|
|
|
all_actions_data$log1p_count <- log1p(all_actions_data$count)
|
2024-05-09 22:05:21 +00:00
|
|
|
# 3 rdd in lmer analysis
|
|
|
|
|
# rdd: https://rpubs.com/phle/r_tutorial_regression_discontinuity_design
|
|
|
|
|
# lmer: https://www.youtube.com/watch?v=LzAwEKrn2Mc
|
2024-04-27 19:30:20 +00:00
|
|
|
library(lme4)
|
2024-05-09 22:05:21 +00:00
|
|
|
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
|
2024-04-27 19:30:20 +00:00
|
|
|
library(optimx)
|
2024-05-07 23:40:38 +00:00
|
|
|
library(lattice)
|
2024-05-09 22:05:21 +00:00
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE, control = lmerControl(
|
2024-04-27 19:30:20 +00:00
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
2024-05-07 23:40:38 +00:00
|
|
|
#identifying the quartiles of effect for D
|
2024-05-09 22:05:21 +00:00
|
|
|
mmcm = coef(all_model)$upstream_vcs_link[, 1]
|
|
|
|
|
vcov.vals = as.data.frame(VarCorr(all_model))
|
|
|
|
|
View(vcov.vals)
|
|
|
|
|
#identifying the quartiles of effect for D
|
|
|
|
|
mmcm = coef(all_model)$upstream_vcs_link
|
|
|
|
|
View(mmcm)
|
|
|
|
|
summary(all_model)$coef[,2]
|
|
|
|
|
View(mmcm)
|
|
|
|
|
variance_components <- VarCorr(all_model)
|
|
|
|
|
group_variance <- attr(variance_components$upstream_vcs_link, "stddev")^2
|
|
|
|
|
View(mmcm)
|
|
|
|
|
fixef(all())
|
|
|
|
|
fixef(all_model
|
|
|
|
|
summary(all_model)$coef[,2]
|
|
|
|
|
fixef(all_model)
|
|
|
|
|
fixed_impacts = fixef(all_model)
|
|
|
|
|
dotplot(all_model_ranef_condvar)
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
dotplot(all_model_ranef_condvar)
|
|
|
|
|
broom.mixed::tidy(all_model, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
test <- broom.mixed::tidy(all_model, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
View(test)
|
|
|
|
|
all_gmodel <- glmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, family = Gamma)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, family = Gamma)
|
|
|
|
|
all_gmodel <- glmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, family=poisson)
|
|
|
|
|
all_gmodel <- glmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link), data=all_actions_data, family=poisson)
|
|
|
|
|
all_gmodel <- glmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link), data=all_actions_data, family=binomial)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link), data=all_actions_data, family=binomial)
|
|
|
|
|
df_ranefs <- as.data.frame(all_model_ranef_condvar)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link), data=all_actions_data, family=binomial)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (1 | upstream_vcs_link), data=all_actions_data, family=poisson)
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_gmodel, condVar = TRUE)
|
|
|
|
|
all_model_ranef_condvar <- ranef(all_model, condVar = TRUE)
|
|
|
|
|
all_gmodel_ranef_condvar <- ranef(all_gmodel, condVar = TRUE)
|
|
|
|
|
View(all_gmodel_ranef_condvar)
|
|
|
|
|
test <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
View(test)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)+ scaled_project_age | upstream_vcs_link), data=all_actions_data)
|
|
|
|
|
test <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
View(test)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=TRUE, control = lmerControl(
|
|
|
|
|
optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
|
|
|
|
|
test <- broom.mixed::tidy(all_model, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
View(test)
|
|
|
|
|
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
View(test_condvals)
|
|
|
|
|
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
|
|
|
|
|
View(test_glmer_ranef_D)
|
|
|
|
|
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
|
|
|
|
|
has_zero <- function(estimate, low, high){
|
|
|
|
|
return(ifelse((low < 0),ifelse((high > 0), 1, 0), 2))
|
2024-05-07 23:40:38 +00:00
|
|
|
}
|
2024-05-09 22:05:21 +00:00
|
|
|
test_glmer_ranef_D <- test_glmer_ranef_D |>
|
|
|
|
|
mutate(ranef_grouping = has_zero(estimate, conf.low, conf.high)) |>
|
|
|
|
|
mutate(rank = rank(estimate))
|
|
|
|
|
test_glmer_ranef_D |>
|
2024-05-08 14:33:03 +00:00
|
|
|
ggplot(aes(x=rank, y=condval, col = as.factor(ranef_grouping))) +
|
2024-05-09 22:05:21 +00:00
|
|
|
geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
|
2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
test_glmer_ranef_D |>
|
|
|
|
|
ggplot(aes(x=rank, y=estimate, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
|
2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
summary(all_gmodel)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
|
|
|
|
|
has_zero <- function(estimate, low, high){
|
|
|
|
|
return(ifelse((low < 0),ifelse((high > 0), 1, 0), 2))
|
|
|
|
|
}
|
|
|
|
|
test_glmer_ranef_D <- test_glmer_ranef_D |>
|
|
|
|
|
mutate(ranef_grouping = has_zero(estimate, conf.low, conf.high)) |>
|
|
|
|
|
mutate(rank = rank(estimate))
|
|
|
|
|
test_glmer_ranef_D |>
|
|
|
|
|
ggplot(aes(x=rank, y=estimate, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
|
2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
View(test_glmer_ranef_D)
|
|
|
|
|
View(test_condvals)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
View(test_condvals)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, family = Poisson)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, family = poisson)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link), data=all_actions_data, family = poisson)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
|
|
|
|
|
has_zero <- function(estimate, low, high){
|
|
|
|
|
return(ifelse((low < 0),ifelse((high > 0), 1, 0), 2))
|
|
|
|
|
}
|
|
|
|
|
test_glmer_ranef_D <- test_glmer_ranef_D |>
|
|
|
|
|
mutate(ranef_grouping = has_zero(estimate, conf.low, conf.high)) |>
|
|
|
|
|
mutate(rank = rank(estimate))
|
|
|
|
|
test_glmer_ranef_D |>
|
|
|
|
|
ggplot(aes(x=rank, y=estimate, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
|
2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, family = poisson)
|
|
|
|
|
all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
|
|
|
|
|
has_zero <- function(estimate, low, high){
|
|
|
|
|
return(ifelse((low < 0),ifelse((high > 0), 1, 0), 2))
|
|
|
|
|
}
|
|
|
|
|
test_glmer_ranef_D <- test_glmer_ranef_D |>
|
|
|
|
|
mutate(ranef_grouping = has_zero(estimate, conf.low, conf.high)) |>
|
|
|
|
|
mutate(rank = rank(estimate))
|
|
|
|
|
test_glmer_ranef_D |>
|
|
|
|
|
ggplot(aes(x=rank, y=estimate, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
|
2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
variance(all_actions_data$log1p_count)
|
|
|
|
|
var(all_actions_data$log1p_count)
|
|
|
|
|
mean (all_actions_data$log1p_count)
|
|
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link),data=all_actions_data)
|
|
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link),
|
|
|
|
|
control=glmerControl(optimizer="bobyqa",
|
|
|
|
|
optCtrl=list(maxfun=2e5)), data=all_actions_data)
|
|
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link),
|
|
|
|
|
control=glmerControl(optimizer="bobyqa",
|
|
|
|
|
optCtrl=list(maxfun=2e5)), data=all_actions_data)
|
|
|
|
|
summary(all_gmodel)
|
|
|
|
|
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
|
|
|
|
|
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
|
|
|
|
|
has_zero <- function(estimate, low, high){
|
|
|
|
|
return(ifelse((low < 0),ifelse((high > 0), 1, 0), 2))
|
|
|
|
|
}
|
|
|
|
|
test_glmer_ranef_D <- test_glmer_ranef_D |>
|
|
|
|
|
mutate(ranef_grouping = has_zero(estimate, conf.low, conf.high)) |>
|
|
|
|
|
mutate(rank = rank(estimate))
|
|
|
|
|
test_glmer_ranef_D |>
|
|
|
|
|
ggplot(aes(x=rank, y=estimate, col = as.factor(ranef_grouping))) +
|
|
|
|
|
geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
|
2024-05-08 14:33:03 +00:00
|
|
|
theme_bw()
|
2024-05-09 22:05:21 +00:00
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
#all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link),
|
|
|
|
|
# control=glmerControl(optimizer="bobyqa",
|
|
|
|
|
# optCtrl=list(maxfun=2e5)), data=all_actions_data)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link), data=all_actions_data)
|
|
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
#all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link),
|
|
|
|
|
# control=glmerControl(optimizer="bobyqa",
|
|
|
|
|
# optCtrl=list(maxfun=2e5)), data=all_actions_data)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link), data=all_actions_data, verbose=TRUE)
|
|
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
#all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link),
|
|
|
|
|
# control=glmerControl(optimizer="bobyqa",
|
|
|
|
|
# optCtrl=list(maxfun=2e5)), data=all_actions_data)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link), data=all_actions_data)
|
|
|
|
|
#all_gmodel <- glmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, nAGQ=0, family = poisson)
|
|
|
|
|
#all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset) | upstream_vcs_link),
|
|
|
|
|
# control=glmerControl(optimizer="bobyqa",
|
|
|
|
|
# optCtrl=list(maxfun=2e5)), data=all_actions_data)
|
|
|
|
|
all_gmodel <- glmer.nb(count ~ D * I(week_offset)+ scaled_project_age + (D | upstream_vcs_link), data=all_actions_data)
|
