# need to calculate inter-class correlation coefficient?
library(merTools)
ICC(outcome="count", group="upstream_vcs_link", data=all_actions_data)
ICC(outcome="count", group="week", data=all_actions_data)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + D * I(week - 26) + age_of_project |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
describe(all_actions_data)
hist(all_actions_data$count)
mean(all_actions_data$count)
median(all_actions_data$count)
mean(all_actions_data$count)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1+week|upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1+D * I(week - 26)|upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1+ upstream_vcs_link|upstream_vcs_link), REML=FALSE, data=all_actions_data)
draft_all_model <- lmer(count ~ (1 | D * I(week - 26) + age_of_project) + (1 |upstream_vcs_link), REML=FALSE, data=all_actions_data)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + I(week - 26) |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + week |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + I(week - 26) |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_all_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
draft_mrg_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), REML=FALSE, data=mrg_actions_data)
summary(draft_mrg_model)
draft_all_model <- lmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
flat_all_model <- lmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project, REML=FALSE, data=all_actions_data)
flat_all_model <- lm(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project, REML=FALSE, data=all_actions_data)
summary(flat_all_model)
draft_all_model <- lmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(draft_all_model)
#find some EDA to identify which types of models might be the best for this
mean(all_actions_data$count)
median(all_actions_data$count)
table(all_actions_data$count)
dist(all_actions_data$count)
var(all_actions_data$count)
sd(all_actions_data$count)
qqplot(all_actions_data$count, all_actions_data$week)
qqnorm(all_actions_data$count)
y <- qunif(ppoints(length(all_actions_data$count)))
qqplot(all_actions_data$count, y)
qqnorm(all_actions_data$count)
qqnorm(log(all_actions_data$count)
qqnorm(log(all_actions_data$count))
qqnorm(log(all_actions_data$count))
qqplot(log(all_actions_data$count), y)
qqnorm(all_actions_data$count)
qqnorm(root(all_actions_data$count))
qqnorm(log(all_actions_data$count))
qqplot(log(all_actions_data$count), y)
qqplot(all_actions_data$count, y)
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
summary(draft_all_model)
# Performance:
draft_mrg_model <- lmer(count ~ D * I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), REML=FALSE, data=mrg_actions_data)
summary(draft_mrg_model)
lmer_residuals <- residuals(lmer_all_model)
lmer_all_model <- lmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(lmer_all_model)
lmer_residuals <- residuals(lmer_all_model)
qqnorm(lmer_residuals)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
summary(poisson_all_model)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"), nAGQ = 100)
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
# this is the file with the lmer multi-level rddAnalysis
library(tidyverse)
library(plyr)
# 0 loading the readme data in
try(setwd(dirname(rstudioapi::getActiveDocumentContext()$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$observation_type == "all"),]
return(longer)
}
expanded_data <- expand_timeseries(readme_df[1,])
for (i in 2:nrow(readme_df)){
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
}
#filter out the windows of time that we're looking at
window_num <- 8
expanded_data <- expanded_data |>
filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
mutate(D = ifelse(week > 26, 1, 0))
#separate out the cleaning d
all_actions_data <- expanded_data[which(expanded_data$observation_type == "all"),]
mrg_actions_data <- expanded_data[which(expanded_data$observation_type == "mrg"),]
#find some EDA to identify which types of models might be the best for this
mean(all_actions_data$count)
median(all_actions_data$count)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"), nAGQ = 100)
summary(poisson_all_model)
# 3 rdd in lmer analysis
# rdd: https://rpubs.com/phle/r_tutorial_regression_discontinuity_design
# lmer: https://www.youtube.com/watch?v=LzAwEKrn2Mc
library(lme4)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"), nAGQ = 100)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"), nAGQ = 100)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + age_of_project + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scale(age_of_project) + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
qqnorm(poisson_residuals)
qqnorm(poisson_residuals)
#scale the age numbers
expanded_data$scaled_project_age <- scale(expanded_data$age_of_project)
#separate out the cleaning d
all_actions_data <- expanded_data[which(expanded_data$observation_type == "all"),]
mrg_actions_data <- expanded_data[which(expanded_data$observation_type == "mrg"),]
#find some EDA to identify which types of models might be the best for this
mean(all_actions_data$count)
median(all_actions_data$count)
table(all_actions_data$count)
var(all_actions_data$count)
qqnorm(all_actions_data$count)
y <- qunif(ppoints(length(all_actions_data$count)))
qqplot(all_actions_data$count, y)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
# 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$after_mrg_ct), ", ")
drop <- c("before_all_ct", "before_mrg_ct", "after_all_ct", "after_mrg_ct")
# 2 some expansion needs to happens for each project
expand_timeseries <- function(project_row) {
longer <- project_row |>
pivot_longer(cols = starts_with("ct"),
names_to = "window",
values_to = "count") |>
unnest(count)
longer$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(readme_df[1,])
for (i in 2:nrow(readme_df)){
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
}
#filter out the windows of time that we're looking at
window_num <- 8
expanded_data <- expanded_data |>
filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
mutate(D = ifelse(week > 26, 1, 0))
# 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)
}
expanded_data <- expand_timeseries(readme_df[1,])
for (i in 2:nrow(readme_df)){
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
}
#filter out the windows of time that we're looking at
window_num <- 8
windowed_data <- expanded_data |>
filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
mutate(D = ifelse(week > 26, 1, 0))
#scale the age numbers
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
#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"),]
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
# 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)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
#logistic regression mixed effects
log_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link),data=all_actions_data, family = binomial)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(scale(count) ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"), control=glmerControl(optimizer="bobyqa"))
#logistic regression mixed effects
log_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link),data=all_actions_data, family = binomial)
qqnorm(poisson_residuals)
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
#logistic regression mixed effects (doesn't work)
log_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + upstream_vcs_link |upstream_vcs_link),data=all_actions_data, family = binomial)
#logistic regression mixed effects (doesn't work)
log_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link),data=all_actions_data, family = binomial)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + upstream_vcs_link |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D ||upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
lmer_all_model <- lmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D ||upstream_vcs_link), REML=FALSE, data=all_actions_data)
summary(lmer_all_model)
lmer_residuals <- residuals(lmer_all_model)
qqnorm(lmer_residuals)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D ||upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
#if I'm reading the residuals right, the poisson is better?
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
# 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)
}
expanded_data <- expand_timeseries(readme_df[1,])
for (i in 2:nrow(readme_df)){
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
}
#filter out the windows of time that we're looking at
window_num <- 8
windowed_data <- expanded_data |>
filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
mutate(D = ifelse(week > 26, 1, 0))
#scale the age numbers
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
#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"),]
#if I'm reading the residuals right, the poisson is better?
# there's a conversation to be had between whether (D |upstream_vcs_link) or (D || upstream_vcs_link)
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_test_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_test_model)
summary(poisson_all_model)
summary(poisson_test_model)
poisson_test_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D ||upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_test_model)
summary(poisson_all_model)
summary(poisson_test_model)
summary(poisson_all_model)
summary(poisson_test_model)
poisson_testing_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 | upstream_vcs_link) +  (0 + D |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_testing_model)
#| label: packages
#| echo: true
library(tidyverse)
library(tidytext)
library(textdata)
library(textstem)
library(tidymodels)
#| label: packages
#| echo: true
library(tidyverse)
library(tidytext)
library(textdata)
library(textstem)
library(tidymodels)
#| label: data 1
#| echo: true
#|
reviews_df <- read_csv("data/rotten_tomatoes_critic_reviews.csv")
reviews_df |> head(2) |> kableExtra::kable()
summary(poisson_test_model)
summary(poisson_all_model)
summary(poisson_all_model)
summary(poisson_test_model)
#if I'm reading the residuals right, the poisson is better?
# there's a conversation to be had between whether (D |upstream_vcs_link) or (D || upstream_vcs_link)
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (week |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
poisson_test_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (week ||upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_test_model)
summary(poisson_all_model)
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (1 + week |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (week |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
ICC(outcome="count", group="week", data=all_actions_data)
library(merTools)
ICC(outcome="count", group="week", data=all_actions_data)
# 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)
}
expanded_data <- expand_timeseries(readme_df[1,])
for (i in 2:nrow(readme_df)){
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
}
#filter out the windows of time that we're looking at
window_num <- 8
windowed_data <- expanded_data |>
filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
mutate(D = ifelse(week > 26, 1, 0))
#scale the age numbers
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
#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"),]
# for visualization, may have to run model for each project and then identify top 5 projects for RDD graphs
#
#
poisson_mrg_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (week |upstream_vcs_link), data=mrg_actions_data, family = poisson(link = "log"))
summary(poisson_mrg_model)
poisson_mrg_residuals <- residuals(poisson_mrg_model)
qqnorm(poisson_mrg_residuals)
# 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)
}
expanded_data <- expand_timeseries(readme_df[1,])
for (i in 2:nrow(readme_df)){
expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
}
#filter out the windows of time that we're looking at
window_num <- 8
windowed_data <- expanded_data |>
filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
mutate(D = ifelse(week > 26, 1, 0))
#scale the age numbers
windowed_data$scaled_project_age <- scale(windowed_data$age_of_project)
#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"),]
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
# (1 |upstream_vcs_link) or (week | upstream_vcs_link)
poisson_all_model <- glmer(count ~ (D + I(week - 26) + D:I(week - 26) + scaled_project_age | upstream_vcs_link) + (1 |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
# 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)
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
# (1 |upstream_vcs_link) or (week | upstream_vcs_link)
poisson_all_model <- glmer(count ~ (D + I(week - 26) + D:I(week - 26) + scaled_project_age | upstream_vcs_link) + (1 |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
# (1 |upstream_vcs_link) or (week | upstream_vcs_link)
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D + I(week - 26) + D:I(week - 26) + scaled_project_age | upstream_vcs_link) + (1 |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
# (1 |upstream_vcs_link) or (week | upstream_vcs_link)
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (D | upstream_vcs_link) + (1 |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
# (D |upstream_vcs_link) or (week | upstream_vcs_link)
poisson_all_model <- glmer(count ~ D + I(week - 26) + D:I(week - 26) + scaled_project_age + (week | upstream_vcs_link) + (1 |upstream_vcs_link), data=all_actions_data, family = poisson(link = "log"))
summary(poisson_all_model)
poisson_residuals <- residuals(poisson_all_model)
qqnorm(poisson_residuals)