GA_analysis.Rmd

---
title: "GA_analysis"
author: "Michael Breshock"
date: "6/25/2021"
output: html_document
editor_options: 
  chunk_output_type: console
---

remember to update

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(googleAnalyticsR)
library(ggplot2)
library(tidyverse)
library(magrittr)
library(ggmap)
library(leaflet)
library(ggpubr)
```

```{r}
theme_set(theme_linedraw() + 
            theme(axis.text = element_text(size = 11, 
                                           color = "black"),
                  axis.title = element_text(size = 12, 
                                            color = "black"),
                  plot.title = element_text(size = 14, 
                                            color = "black",
                                            hjust = .5))
          )
```

```{r sessionsdaily}
ocs_property_id = 256923228
dates = ga_data(ocs_property_id,
        date_range = c("2021-01-01", "yesterday"),
        metric = c("sessions"),
        dimensions = c("date"), limit = 400)
dates
```


```{r devicecategory}
ocs_property_id = 256923228
ses_wk_ctg = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("deviceCategory"))
ses_wk_ctg %<>% mutate(percent = round(100*sessions/sum(sessions), 2),
                       total = sum(sessions)) 

ggplot(ses_wk_ctg, aes(x = as.factor(deviceCategory), y = sessions)) +
  geom_bar(stat = "identity")
```
[Paramaters:](https://developers.google.com/analytics/devguides/reporting/data/v1/api-schema)

sessions: The number of sessions that began on your site or app (event triggered: session_start).

deviceCategory: The type of device: Desktop, Tablet, or Mobile.


```{r sources}
user_sources = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("newUsers"),
          dimensions = c("firstUserSource")) %>%
  subset(grepl("127",firstUserSource) == FALSE) %>% 
  mutate(firstUserSource = as_factor(firstUserSource),
         firstUserSource = fct_infreq(firstUserSource))

ggplot(user_sources, aes(x = firstUserSource, y = newUsers)) +
  geom_bar(stat = "identity") +   theme(axis.text.x=element_text(angle=90,hjust=1,vjust=0.5))
```
(t.co = twitter)

firstUserSource = The source that first acquired the user to your website or app.

newUsers: The number of users who interacted with your site or launched your app for the first time (event triggered: first_open or first_visit).


```{r usersweek}
users_wk = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("totalUsers", "newUsers", "eventCount",
                      "active7DayUsers", "userEngagementDuration"),
          dimensions = c("week")) %>% 
  mutate(week = as.numeric(week),
         avgEngagementDuration = # avg engagement in minutes
           round(userEngagementDuration/(60*totalUsers)))

pivot_longer(users_wk, c(totalUsers, newUsers, active7DayUsers), 
             names_to = "metric", values_to = "count") %>%
  ggplot(aes(x = week, y = count)) +
  geom_line(aes(color = metric))
```

```{r usersday}
ocs_property_id = 256923228
users_date = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("totalUsers", "newUsers",
                      "active1DayUsers"),
          dimensions = c("date"), limit = 1000) %>% 
  pivot_longer(c(newUsers, active1DayUsers), 
              names_to = "metric", values_to = "count")

  ggplot(users_date, aes(x = date, y = count, color = metric)) +
    geom_line() + #scale_color_viridis_d() + 
    labs(x = "Day", y = "User Count", title = "Open Case Studies Daily New and Active Users")
```

```{r usertotals}
#users totals
ocs_property_id = 256923228
total_users = ga_data(ocs_property_id,
        date_range = c("2021-01-01", "yesterday"),
        metrics = c("totalUsers", "newUsers",
                 "active1DayUsers"))
avg_newusers = total_users$newUsers / nrow(dates)
avg_totusers = total_users$active1DayUsers / nrow(dates)

```


```{r avgengage}
ggplot(users_date, aes(x = as.factor(avgEngagementDuration), 
                       fill = as.factor(avgEngagementDuration))) + 
  geom_bar() + scale_fill_viridis_d() + guides(fill = "none") +
  labs(x = "Average Engagement Duration (Minutes)", y = "Number of Days", 
       title = "Count of Daily Average Engagement") 
```


week: The week of the event, a two-digit number from 01 to 53. Each week starts on Sunday. January 1st is always in week 01. The first and last week of the year have fewer than 7 days in most years. Weeks other than the first and the last week of the year always have 7 days. For years where January 1st is a Sunday, the first week of that year and the last week of the prior year have 7 days.

totalUsers: The number of distinct users who have logged at least one event, regardless of whether the site or app was in use when that event was logged.

eventCount: The count of events.

active7DayUsers: The number of distinct active users on your site or app within a 7 day period. The 7 day period includes the last day in the report's date range.

userEngagementDuration: The total amount of time (in seconds) your website or app was in the foreground of users' devices. 


```{r cities}
cities = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("country", "city"), limit = 1000)
cities %<>% mutate(
  latitude = geocode(paste(city, country, sep = ", "))$lat,
  longitude = geocode(paste(city, country, sep = ", "))$lon
)
cities
```

city: The city from which the user activity originated.

```{r}
worldmap = map_data("world")
cities = drop_na(cities)
save(cities, file = "city_coordinates.rda")
```

```{r bubblemap}
# Create a color palette with handmade bins.
mybins <- c(1, 5, 10, 50, 100, 1000, 2000)
mypalette <- colorBin(palette="Reds", domain = cities$sessions,
                      na.color="transparent", bins=mybins)


# Final Map
m <- leaflet(cities) %>% 
  addTiles()  %>% 
  setView(lat = 0, lng = 0, zoom = 1) %>%
  addProviderTiles(providers$CartoDB.Positron) %>%
  addCircleMarkers(~longitude, ~latitude, 
    fillColor = ~mypalette(sessions), fillOpacity = 0.7, radius= ~sqrt(sessions), stroke=FALSE
  ) %>%
  addLegend(pal=mypalette, values=~sessions, opacity=0.9, title = "Sessions", position = "topright" )

m 
```

```{r topcities}
cities10 = drop_na(cities) %>% filter(country != "Peru") %>% head(10) %>% 
  mutate(city = case_when(
    city == "(not set)" ~ " ",
    TRUE ~ city)) %>%
  unite(location, c(city, country), sep = "\n")

ggplot(cities10, aes(x = as_factor(location), y = sessions, fill = as_factor(location))) +
  geom_col() + scale_fill_viridis_d() + 
  labs(x = "City, Country", y = "No. of Sessions", 
       title = "Top Ten Cities With Most Open Case Studies Sessions") + 
  guides(fill = "none") + 
  theme(axis.text.x = element_text(angle = 60, vjust = 0.5)) + 
  scale_y_continuous() + 
  geom_text(aes(label = paste0(sessions)), vjust = -.5) + ylim(c(0,1150))
```


```{r, browser}
ocs_property_id = 256923228
browsers = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("browser"))
head(browsers)
```

```{r, dhm}
ocs_property_id = 256923228
dhm = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("hour"), limit = 1000)
dhm
```

```{r, mediums}
ocs_property_id = 256923228
mediums = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("medium"))
mediums
```

```{r url}
ocs_property_id = 256923228
urls = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("fullPageUrl"))

case_studies = c("ocs-bp-rural-and-urban-obesity", 
                 "ocs-bp-air-pollution",
                 "ocs-bp-vaping-case-study",
                 "ocs-bp-opioid-rural-urban",
                 "ocs-bp-RTC-wrangling", 
                 "ocs-bp-RTC-analysis", 
                 "ocs-bp-youth-disconnection",
                 "ocs-bp-youth-mental-health", 
                 "ocs-bp-school-shootings-dashboard",
                 "ocs-bp-co2-emissions",
                 "ocs-bp-diet")

pageUrl = urls %>% mutate(
  interactive = str_detect(fullPageUrl, "rsconnect")) %>% 
  filter(str_detect(fullPageUrl, "opencasestudies.org") | 
           str_detect(fullPageUrl, "jhsph.edu/ocs")) %>%
  mutate(
    code = case_when(
      str_detect(fullPageUrl, "ocs-bp-school-shootings-dashboard") == TRUE ~ "School Shootings", 
      str_detect(fullPageUrl, "ocs-bp-youth-disconnection") == TRUE ~ "Youth Disconnection", 
      str_detect(fullPageUrl, "ocs-bp-opioid-rural-urban") == TRUE ~ "Opiods",
      str_detect(fullPageUrl, "ocs-bp-vaping-case-study") == TRUE ~ "Vaping",
      str_detect(fullPageUrl, "ocs-bp-youth-mental-health") == TRUE ~ "Mental Health",
      str_detect(fullPageUrl, "ocs-bp-rural-and-urban-obesity") == TRUE ~ "Obesity", 
      str_detect(fullPageUrl, "ocs-bp-RTC-analysis") == TRUE ~ "RTC Analysis",
      str_detect(fullPageUrl, "ocs-bp-co2-emissions") == TRUE ~ "CO2 Emissions",
      str_detect(fullPageUrl, "ocs-bp-RTC-wrangling") == TRUE ~ "RTC Wrangling",
      str_detect(fullPageUrl, "ocs-bp-diet") == TRUE ~ "Diet",
      str_detect(fullPageUrl, "ocs-bp-air-pollution") == TRUE ~ "Air Pollution"
    )
  ) 

pageTotals = pageUrl %>% group_by(code) %>% 
  mutate(ocs_sum = sum(sessions)) %>% select(code, ocs_sum) %>% distinct()

interactiveTotals = pageUrl %>% filter(interactive == TRUE) %>% group_by(code) %>%
  mutate(ocs_sum = sum(sessions)) %>% select(code, ocs_sum, interactive) %>% distinct()

staticTotals = pageUrl %>% filter(interactive == FALSE) %>% group_by(code) %>%
  mutate(ocs_sum = sum(sessions)) %>% select(code, ocs_sum, interactive) %>% distinct()

cstotalsplot = ggplot(drop_na(pageTotals), aes(x = reorder(code, -ocs_sum), y = ocs_sum, fill = code)) +
  geom_bar(position = "dodge", stat = "identity") + scale_fill_viridis_d() + 
  labs(x = "Case Study", y = "Session Count", 
       title = "Number of Total Sessions by Case Study") + 
  theme(axis.text.x = element_text(angle = 60, vjust = 0.5)) + 
  guides(fill = "none") + geom_text(aes(label = paste0(ocs_sum)), vjust = -.5)
```

```{r staticeplot}
staticplot = ggplot(drop_na(staticTotals), 
                          aes(x = reorder(code, -ocs_sum), y = ocs_sum, fill = code)) +
  geom_bar(position = "dodge", stat = "identity") + scale_fill_viridis_d() + 
  labs(x = "Case Study", y = "Session Count", 
       title = "Number of Total Sessions by Static Case Study") + 
  theme(axis.text.x = element_text(angle = 60, vjust = 0.5)) + 
  guides(fill = "none") + geom_text(aes(label = paste0(ocs_sum)), vjust = -.5) +
  ylim(0,1150)
staticplot
```

```{r interactiveplot}
interactivesplot = ggplot(drop_na(interactiveTotals), 
                          aes(x = reorder(code, -ocs_sum), y = ocs_sum, fill = code)) +
  geom_bar(position = "dodge", stat = "identity") + scale_fill_viridis_d() + 
  labs(x = "Case Study", y = "Session Count", 
       title = "Number of Total Sessions by Interactive Case Study") + 
  theme(axis.text.x = element_text(angle = 60, vjust = 0.5)) + 
  guides(fill = "none") + geom_text(aes(label = paste0(ocs_sum)), vjust = -.5) + 
  ylim(0,500)
interactivesplot
```

```{r case_study_sessionsplot}
ggarrange(staticplot, interactivesplot,
          labels = c("A.", "B."), 
          ncol = 1, nrow = 2)
```


```{r langauge}
ocs_property_id = 256923228
language = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("newUsers"),
          dimensions = c("language"))
language
```

```{r, linkUrl}
ocs_property_id = 256923228
linkUrl = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("linkUrl"))
linkUrl
```

```{r, OSversion}
ocs_property_id = 256923228
OSversion = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("operatingSystem"))
OSversion
```

```{r, pageReferrer}
ocs_property_id = 256923228
pageReferrer = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("pageReferrer"))
pageReferrer
```


look into collapsing duplicates functions / group_by!
```{r pagetitle}
ocs_property_id = 256923228
# used fullPageUrl instead 
# pagetitle = ga_data(ocs_property_id,
#           date_range = c("2021-01-01", "yesterday"),
#           metrics = c("sessions"),
#           dimensions = c("pageTitle"))
# case_studies = c("School Shootings in the United States", 
#                  "Disparities in Youth Disconnection",
#                  "Opioids in United States",
#                  "Vaping Behaviors in American Youth",
#                  "Mental Health of American Youth", 
#                  "Exploring global patterns of obesity across rural and urban regions",  "Influence of Multicollinearity on Measured Impact of Right-to-Carry Gun Laws", 
#                  "Exploring CO2 emissions across time",
#                  "Influence of Multicollinearity on Measured Impact of Right-to-Carry Gun Laws Part 1", 
#                  "Exploring global patterns of dietary behaviors associated with health risk",
#                  "Predicting Annual Air Pollution")
# pages = pagetitle %>% mutate(
#   ocs = str_remove(pageTitle, pattern = "Open Case Studies: ") 
# ) %>% 
#   filter(ocs %in% case_studies) %>% 
#   mutate(
#     static = str_count(pageTitle, "Open Case Studies: ")) %>%
#   mutate(
#     code = case_when(
#       ocs == "School Shootings in the United States" ~ "School Shootings", 
#       ocs == "Disparities in Youth Disconnection" ~ "Youth Disconnection", 
#       ocs == "Opioids in United States" ~ "Opiods",
#       ocs == "Vaping Behaviors in American Youth" ~ "Vaping",
#       ocs == "Mental Health of American Youth" ~ "Mental Health",
#       ocs == "Exploring global patterns of obesity across rural and urban regions" ~ "Obesity", 
#       ocs == "Influence of Multicollinearity on Measured Impact of Right-to-Carry Gun Laws" ~ "RTC Analysis",
#       ocs == "Exploring CO2 emissions across time" ~ "CO2 Emissions",
#       ocs == "Influence of Multicollinearity on Measured Impact of Right-to-Carry Gun Laws Part 1" ~ "RTC Wrangling",
#       ocs == "Exploring global patterns of dietary behaviors associated with health risk" ~ "Diet",
#       ocs == "Predicting Annual Air Pollution" ~ "Air Pollution"
#     )
#   ) %>% group_by(code) %>% 
#   mutate(ocs_sum = sum(sessions)) %>% select(code, ocs_sum) %>% distinct()
# pages
```

```{r percentScrolled}
ocs_property_id = 256923228
percentScrolled = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("percentScrolled"))
percentScrolled
```

```{r searchTerm}
ocs_property_id = 256923228
searchTerm = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("searchTerm"))
searchTerm
```

```{r sources}
ocs_property_id = 256923228
sources = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessions"),
          dimensions = c("source"))
sources
```

```{r sessions}
ocs_property_id = 256923228
engagedSessions = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("engagedSessions", "engagementRate", "sessions"),
          dimensions = c("week"), limit = 1000)

ses_long = engagedSessions %>%
  pivot_longer(c(sessions, engagedSessions), names_to = "metric", values_to = "count")

sesplot = ggplot(ses_long, aes(x = as.numeric(week), y = count, color = metric)) +
  geom_line() +
  labs(x = "Week Number", y = "Count", title = "Open Case Studies Weekly Sessions and Engaged Sessions")
```

```{r engagedSessions}
engageplot = ggplot(engagedSessions, aes(x = as.numeric(week), y = engagementRate, 
                            color = engagementRate)) +
  geom_line() + 
  labs(x = "Week Number", y = "Engagement Rate", 
       title = "Open Case Studies Weekly Engagement Rate") + guides(color = "none") 
  #scale_color_manual(labels = "engagmentRate", values = "#8C4799")
```

```{r engagementplot}
ggarrange(sesplot, engageplot,
          labels = c("A.", "B."), 
          ncol = 1, nrow = 2)
```


```{r sessiontotals}
#totals
totalSessions = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "today"),
          metrics = c("engagedSessions", "engagementRate", "sessions"),
          limit = 1000)
avgsessions = totalSessions$sessions / nrow(dates)
avgengagedsessions = totalSessions$engagedSessions / nrow(dates)
```


```{r, screenPageViews}
ocs_property_id = 256923228
screenPageViews = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("screenPageViews"),
          dimensions = c("pageTitle"), limit = 1000)
screenPageViews
```

```{r, sessionsPerUser}
ocs_property_id = 256923228
sessionsPerUser = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("sessionsPerUser"),
          dimensions = c("pageTitle"), limit = 1000)
sessionsPerUser
```

```{r, conversions}
ocs_property_id = 256923228
conversions = ga_data(ocs_property_id,
          date_range = c("2021-01-01", "yesterday"),
          metrics = c("conversions"),
          dimensions = c("eventName"), limit = 1000)
conversions
```


dimensions of interest:
- browser
- cohort
- country/countryID
- dateHourMinute
- firstSessionDate
- fileName
- firstUserMedium 
- firstUserTrafficOrigin
- fullPageUrl
- hour, minute
- language, languageCode
- linkClasses, linkDomain, linkUrl
- medium, method
- operatingSystem, operatingSystemWithVersion
- pageLocation, pageReferrer
- pageTitle
- percentScrolled
- searchTerm
- streamName
- sessionSource, source, sessionMedium

metrics of interest: 
- active1DayUsers
- conversions
- engagedSessions, engagementRate
- screenPageViews
- sessions, sessionsPerUser

for users data: use active1Dayusers and newUsers

 
cool figure to make: # of events (or users) per day (events y axis, days x axis)
have a timeline of project development along the x-axis to view if website traffic was changed at all by project developments. 


helpful links: 
https://www.dase-analytics.com/blog/en/how-to-get-data-from-google-analytics-using-r-programming/

https://ga-dev-tools.web.app/dimensions-metrics-explorer/

https://developers.google.com/analytics/devguides/reporting/data/v1/api-schema