From 73d5c5b979809d2be0c394464d4549199f4db469 Mon Sep 17 00:00:00 2001
From: mjgaughan <mgaughan@proton.me>
Date: Wed, 24 Apr 2024 16:55:40 -0500
Subject: [PATCH] fixing optimizer for model convergance

---
 R/.Rhistory           | 212 +++++++++++++++++++++---------------------
 R/readmeRDDAnalysis.R |   5 +-
 2 files changed, 110 insertions(+), 107 deletions(-)

diff --git a/R/.Rhistory b/R/.Rhistory
index 18b8514..74a2e43 100644
--- a/R/.Rhistory
+++ b/R/.Rhistory
@@ -1,109 +1,3 @@
-p
-# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
-# (D |upstream_vcs_link) or (week | upstream_vcs_link)
-test_model <- lmer(count ~ D * I(week_offset) + scaled_project_age + (D * I(week_offset) |upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw()
-p
-#test model
-test_model <- lmer(count ~ D * I(week_offset) + scaled_project_age + (D |upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
-summary(test_model)
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw()
-p
-#test model
-test_model <- lmer(count ~ D * I(week_offset) + scaled_project_age + (D * I(week_offset)|upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
-summary(test_model)
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw()
-p
-# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
-#making some random data
-sampled_data <- readme_df[sample(nrow(readme_df), 8), ]
-expanded_sample_data <- expand_timeseries(sampled_data[1,])
-for (i in 2:nrow(sampled_data)){
-expanded_sample_data <- rbind(expanded_sample_data, expand_timeseries(sampled_data[i,]))
-}
-windowed_sample_data <- expanded_sample_data |>
-filter(week >= (26 - window_num) & week <= (26 + window_num)) |>
-mutate(D = ifelse(week > 26, 1, 0))
-windowed_sample_data$scaled_project_age <- scale(windowed_sample_data$age_of_project)
-windowed_sample_data$week_offset <- windowed_sample_data$week - 26
-all_actions_sample_data <- windowed_sample_data[which(windowed_sample_data$observation_type == "all"),]
-#test model
-test_model <- lmer(count ~ D * I(week_offset) + scaled_project_age + (D * I(week_offset)|upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
-summary(test_model)
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw()
-p
-#test model
-test_model <- lmer(count ~ D * I(week_offset) + scaled_project_age + (week_offset|upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
-summary(test_model)
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw()
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-guides(fill="none") +
-theme_bw()
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme(legend.position="none") +
-theme_bw()
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link)) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw(legend.position="none")
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link), show.legend = FALSE) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw(legend.position="none")
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link), show.legend = FALSE) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link)) +
-theme_bw()
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link), show.legend = FALSE) +
-geom_point(size=3) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link), show.legend = FALSE) +
-theme_bw()
-p
-#plot results
-p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link), show.legend = FALSE) +
-geom_point(size=3, show.legend = FALSE) +
-geom_line(aes(y=predict(test_model), group=upstream_vcs_link), show.legend = FALSE) +
-theme_bw()
-p
-# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
-#making some random data
-sampled_data <- readme_df[sample(nrow(readme_df), 220), ]
 expanded_sample_data <- expand_timeseries(sampled_data[1,])
 for (i in 2:nrow(sampled_data)){
 expanded_sample_data <- rbind(expanded_sample_data, expand_timeseries(sampled_data[i,]))
@@ -510,3 +404,109 @@ summary(all_model)
 ##end of the model testing and plotting section
 all_model <- lmer(count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE)
 summary(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$observation_type == "all"),]
+return(longer)
+}
+expanded_data <- expand_timeseries(readme_df[1,])
+for (i in 2:nrow(readme_df)){
+expanded_data <- rbind(expanded_data, expand_timeseries(readme_df[i,]))
+}
+#filter out the windows of time that we're looking at
+window_num <- 8
+windowed_data <- expanded_data |>
+filter(week >= (27 - window_num) & week <= (27 + window_num)) |>
+mutate(D = ifelse(week > 27, 1, 0))
+#scale the age numbers
+windowed_data$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$observation_type == "mrg"),]
+#find some EDA to identify which types of models might be the best for this
+hist(all_actions_data$count)
+#find some EDA to identify which types of models might be the best for this
+hist(log1p(all_actions_data$count))
+#find some EDA to identify which types of models might be the best for this
+hist(log(all_actions_data$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)
+##end of the model testing and plotting section
+all_model <- lmer(log(count) ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE)
+all_actions_data$logged_count <- log(all_actions_data$count)
+all_actions_data$log1p_count <- log1p(all_actions_data$count)
+##end of the model testing and plotting section
+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)
+summary(all_model)
+all_residuals <- residuals(all_model)
+qqnorm(all_residuals)
+##end of the model testing and plotting section
+all_model <- lmer(logged_count ~ D * I(week_offset)+ scaled_project_age + (D * I(week_offset)| upstream_vcs_link), data=all_actions_data, REML=FALSE)
+##end of the model testing and plotting section
+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)
+##end of the model testing and plotting section
+all_model <- lmer(log1p_count ~ D * I(week_offset)+ scaled_project_age + (week_offset| upstream_vcs_link), data=all_actions_data, REML=FALSE)
+##end of the model testing and plotting section
+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)
+# https://www.bristol.ac.uk/cmm/learning/videos/random-intercepts.html#exvar
+#making some random data
+sampled_data <- readme_df[sample(nrow(readme_df), 220), ]
+expanded_sample_data <- expand_timeseries(sampled_data[1,])
+for (i in 2:nrow(sampled_data)){
+expanded_sample_data <- rbind(expanded_sample_data, expand_timeseries(sampled_data[i,]))
+}
+windowed_sample_data <- expanded_sample_data |>
+filter(week >= (27 - window_num) & week <= (27 + window_num)) |>
+mutate(D = ifelse(week > 27, 1, 0))
+windowed_sample_data$scaled_project_age <- scale(windowed_sample_data$age_of_project)
+windowed_sample_data$week_offset <- windowed_sample_data$week - 27
+all_actions_sample_data <- windowed_sample_data[which(windowed_sample_data$observation_type == "all"),]
+#test model
+test_model <- lmer(log1p_count ~ D * I(week_offset) + scaled_project_age + (D * I(week_offset)|upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
+all_actions_sample_data$log1p_count <- log1p(all_actions_sample_data$count)
+#test model
+test_model <- lmer(log1p_count ~ D * I(week_offset) + scaled_project_age + (D * I(week_offset)|upstream_vcs_link),  data=all_actions_sample_data, REML=FALSE)
+summary(test_model)
+#plot results
+p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link), show.legend = FALSE) +
+geom_point(size=3, show.legend = FALSE) +
+geom_line(aes(y=predict(test_model)), show.legend = FALSE) +
+theme_bw()
+p
+#plot results
+p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_vcs_link), show.legend = FALSE) +
+geom_point(size=3, show.legend = FALSE) +
+geom_line(aes(y=predict(test_model)), show.legend = FALSE) +
+theme_bw()
+p
+library(merTools)
+ICC(outcome="count", group="week", data=all_actions_data)
+ICC(outcome="count", group="upstream_vcs_link", data=all_actions_data)
+ICC(outcome="count", group="week", data=all_actions_data)
diff --git a/R/readmeRDDAnalysis.R b/R/readmeRDDAnalysis.R
index 84200e5..67ba780 100644
--- a/R/readmeRDDAnalysis.R
+++ b/R/readmeRDDAnalysis.R
@@ -80,8 +80,11 @@ p <- ggplot(all_actions_sample_data, aes(x=week_offset, y=count, color=upstream_
   theme_bw() 
 p
 ##end of the model testing and plotting section
-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)
+library(optimx)
+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')))
 summary(all_model)
+random_effects <- ranef(all_model)
 all_residuals <- residuals(all_model)
 qqnorm(all_residuals)
 # for visualization, may have to run model for each project and then identify top 5 projects for RDD graphs