24_deb_pkg_gov/R/contribRDDAnalysis.R

library(tidyverse)
library(plyr) 
#get the contrib data instead
try(setwd(dirname(rstudioapi::getActiveDocumentContext()$path)))
contrib_df <- read_csv("../final_data/deb_contrib_did.csv")
#some preprocessing and expansion 
col_order <- c("upstream_vcs_link", "age_of_project", "event_date", "event_hash", "before_all_ct", "after_all_ct", "before_mrg_ct", "after_mrg_ct", "before_auth_new", "after_auth_new", "before_commit_new",  "after_commit_new")
contrib_df <- contrib_df[,col_order]
contrib_df$ct_before_all <- str_split(gsub("[][]","", contrib_df$before_all_ct), ", ")
contrib_df$ct_after_all <- str_split(gsub("[][]","", contrib_df$after_all_ct), ", ")
contrib_df$ct_before_mrg <- str_split(gsub("[][]","", contrib_df$before_mrg_ct), ", ")
contrib_df$ct_after_mrg <- str_split(gsub("[][]","", contrib_df$after_mrg_ct), ", ")
drop <- c("before_all_ct", "before_mrg_ct", "after_all_ct", "after_mrg_ct")
contrib_df = contrib_df[,!(names(contrib_df) %in% drop)]
# 2 some expansion needs to happens for each project 
expand_timeseries <- function(project_row) {
  longer <- project_row |> 
    pivot_longer(cols = starts_with("ct"),
                 names_to = "window",
                 values_to = "count") |>
    unnest(count) 
  longer$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)
}
expanded_data <- expand_timeseries(contrib_df[1,])
for (i in 2:nrow(contrib_df)){
  expanded_data <- rbind(expanded_data, expand_timeseries(contrib_df[i,]))
}
#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 and calculate the week offset here
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
windowed_data$week_offset <- windowed_data$week - 27
#break out the different type of commit actions
all_actions_data <- windowed_data[which(windowed_data$observation_type == "all"),]
mrg_actions_data <- windowed_data[which(windowed_data$observation_type == "mrg"),]
#logging
all_actions_data$logged_count <- log(all_actions_data$count)
all_actions_data$log1p_count <- log1p(all_actions_data$count)
#EDA
range(all_actions_data$log1p_count) # 0.000000 6.745236
mean(all_actions_data$log1p_count) # 1.200043
var(all_actions_data$log1p_count) # 1.753764
median(all_actions_data$log1p_count) # 0.6931472
# now for merge
mrg_actions_data$logged_count <- log(mrg_actions_data$count)
mrg_actions_data$log1p_count <- log1p(mrg_actions_data$count)
#imports for models 
library(lme4)
library(optimx)
library(lattice)
#model
all_gmodel <- glmer.nb(log1p_count ~ D * week_offset + scaled_project_age + (D * week_offset | upstream_vcs_link),
              control=glmerControl(optimizer="bobyqa",
                                           optCtrl=list(maxfun=2e5)), nAGQ=0, data=all_actions_data)
summary(all_gmodel)
saveRDS(all_gmodel, "0512_contrib_all.rda")
all_residuals <- residuals(all_gmodel)
qqnorm(all_residuals)
#identifying the quartiles of effect for D
test_condvals <- broom.mixed::tidy(all_gmodel, effects = "ran_vals", conf.int = TRUE)
test_glmer_ranef_D <- test_condvals [which(test_condvals $term == "D"),]
has_zero <- function(estimate, low, high){
  return(ifelse((low < 0),ifelse((high > 0), 1, 0), 2))
}
test_glmer_ranef_D <- test_glmer_ranef_D |>
  mutate(ranef_grouping = has_zero(estimate, conf.low, conf.high)) |>
  mutate(rank = rank(estimate))
g <- test_glmer_ranef_D |> 
  ggplot(aes(x=rank, y=estimate, col = as.factor(ranef_grouping))) +
  geom_linerange(aes(ymin= conf.low, ymax= conf.high)) +
  theme_bw()
g
write.csv(test_glmer_ranef_D, "051224_contrib_grouped.csv")
#NOTE: The merge action model below this has not been used but this is what it would be if it was
mrg_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (week_offset | upstream_vcs_link), data=mrg_actions_data, REML=FALSE, control = lmerControl(
  optimizer ='optimx', optCtrl=list(method='L-BFGS-B')))
summary(mrg_model)
#identifying the quartiles of effect for D
mrg_model_ranef <- ranef(mrg_model)
dotplot(mrg_model_ranef)
d_effect_ranef_mrg <- mrg_model_ranef[mrg_model_ranef$term=="D",]
d_effect_ranef_mrg$quartile <- ntile(d_effect_ranef_mrg$condval, 4)
#merge model residuals
mrg_residuals <- residuals(mrg_model)
qqnorm(mrg_residuals)