overall_analysis.qmd

# Overall analysis

This section covers all workshops combined.

## Raw data sources
This consists of four datasets.
- Registration numbers (pre-workshop interest) from LibCal
  - Downloaded per calendar then concatenated (?)
  - Registered <- read.csv("raw_data/processed_Registered.csv")
- Attendance from LibInsights for Course-integrated instruction
  - How to get this data via download from "Course Integrated Library Instruction"
    - Method 1
      - Filter by Custom Date Range (July 1, 2018, to present)
      - Add Additional Report Filter "Name of Workshop" "Is Not" "Null", then click "Add this filter".
      - Click "generate report".
    - Method 2
      - Use saved filter "Workshop Filters (CMC)"
    - Method 2
      - Use saved filter "Workshops in Courses (CMC)"
    - After either method, click "Reports" tab and then green button "Export Data to CSV"
    - Save the file in OSF as /raw_data/attendance_courses.csv
- Attendance from LibInsights for Outreach
  - How to get this data via download from "Outreach and Programming Form Analysis"
    - Method 1
      - Filter by Custom Date Range (July 1, 2018, to present)
      - Add Additional Report Filter "Type of Outreach" "Is" "Workshop", then click "Add this filter".
      - Click "generate report".
    - Method 2
      - Use saved filter "Workshop Filters (CMC)"
  - After either method, click "Reports" tab and then green button "Export Data to CSV"
    - Save the file in OSF as /raw_data/attendance_outreach.csv
- Post-workshop surveys (feedback and satisfaction) from LibWizard
  - postworkshopsurveys_named <- read.csv("raw_data/processed_postworkshopsurveys_named.csv")

## Data download from OSF

```{r}
# Data for this report are gathered from three sources: pre-registration questionnaire when people register, attendance numbers entered by workshop instructors in LibInsights, and post-workshop surveys.  Registration numbers include learner names and emails and so all three datasets are kept on a private OSF repository and must be downloaded there.  Do not include these in github.

###########
# Load data from OSF
###########

# https://www.statology.org/r-check-if-file-exists/

# Authenticate
# Your OSF PAT should be in .REnviron file following these instructions:
# https://docs.ropensci.org/osfr/reference/osf_auth.html

if(file.exists("raw_data/README.md")){
  print("OSF data already downloaded.")
} else {
  print("Download it!")

library(osfr)

# ## Retrieve project
osf_node <- osf_retrieve_node("qjkvf")

## List files in project
osf_files <- osf_ls_files(osf_node)

## Download the files
osf_download(x = osf_files,
              path = NULL, #default save to local working directory
              recurse = TRUE, #download all nested files
              conflicts = "overwrite", #OSF is the canonical version
              progress = TRUE #show progress bar
 )
}

```



## Data processing


Titles have varied over time with marketing experiments, typos, and accidental changes.  This code assigns a standardized code between each topic.

```{r}
library(dplyr)


topics_complete_variations <- read.csv("./workshop_metadata.csv")

# remove any accidental duplicates of name variations (otherwise will create a many-to-many join below in next step)
topics_complete_variations_cleaned <-topics_complete_variations %>%
  dplyr::distinct(WorkshopNameVariations, WorkshopCode)

# Must attach standard titles to all three datasets (registrations, attendance, and post-workshop surveys) to allow joining the three based on topic and date.

```


The four raw datasets are combined and cleaned into three working datasets.

- Registration numbers

- Attendance
  
  - Course and Outreach datasets are each given a column for Is.Course (Y = Course, N = Outreach) before being combined.

```{r}
library(lubridate)

###########
# Attendance data is created by combining Course Instruction attendance data from LibInsights and Outreach Instruction attendance data from LibInsights.  It can be filtered by columns such as date, year, month, instructor, and workshop topic.
###########

# Outreach
attendance_outreach <- read.csv("raw_data/attendance_outreach.csv",
                                header=TRUE,
                                stringsAsFactors=FALSE)
attendance_outreach$Is.Course <- "Not a course"
attendance_outreach$Requested.[is.na(attendance_outreach$Requested.)] <- "Unknown"
attendance_outreach$Requested.[attendance_outreach$Requested.==""] <- "Unknown"

# Courses
attendance_courses <- read.csv("raw_data/attendance_courses.csv",
                                header=TRUE,
                                stringsAsFactors=FALSE)
attendance_courses$Is.Course <- "Within a course"
attendance_courses$Requested. <- "Yes"
attendance_courses$Total.Attendees <- attendance_courses$Total.Attendance


###########
# We clean dates out by weekday, year, month, day, and times for later analysis and data subsetting.
###########

# Outreach

# Month, day, year, hour, minute, seconds
attendance_outreach$Event.Date.and.Time_2 <- ymd_hms(attendance_outreach$Event.Date.and.Time,
                                                    truncated = 3)


# separate year, month, date, times
attendance_outreach$year <- year(attendance_outreach$Event.Date.and.Time_2)
attendance_outreach$month <- month(attendance_outreach$Event.Date.and.Time_2)
attendance_outreach$day <- day(attendance_outreach$Event.Date.and.Time_2)
attendance_outreach$wday <- wday(attendance_outreach$Event.Date.and.Time_2,
                                      week_start = 1, #1 = Monday,
                                      label = TRUE    # days of weeks as characters
                                      )
attendance_outreach$hour <-as.numeric(
  format(attendance_outreach$Event.Date.and.Time_2, "%H"))

attendance_outreach$Date <-
  format.Date(attendance_outreach$Event.Date.and.Time_2, "%m/%d/%Y")

# convert 00 hour to NA
# attendance_outreach[attendance_outreach$hour==00, "hour"] <- NA

attendance_outreach$hour <- as.numeric(attendance_outreach$hour)

# Semester defined as spring (month 1-5), summer (month 6-7), fall (month 8-12)

attendance_outreach$Semester[attendance_outreach$month>0&
                            attendance_outreach$month<6] <- "Spring"
attendance_outreach$Semester[attendance_outreach$month>5&
                            attendance_outreach$month<8] <- "Spring"
attendance_outreach$Semester[attendance_outreach$month>7&
                            attendance_outreach$month<=12] <- "Spring"


# Courses
# Month, day, year, hour, minute, seconds
attendance_courses$Event.Date.and.Time_2 <-
  ymd_hms(attendance_courses$Session.Date,
          truncated = 3)


# separate year, month, date, times
attendance_courses$year <- year(attendance_courses$Event.Date.and.Time_2)
attendance_courses$month <- month(attendance_courses$Event.Date.and.Time_2)
attendance_courses$day <- day(attendance_courses$Event.Date.and.Time_2)
attendance_courses$wday <- wday(attendance_courses$Event.Date.and.Time_2,
                                      week_start = 1, #1 = Monday,
                                      label = TRUE    # days of weeks as characters
                                      )
attendance_courses$hour <-as.numeric(
  format(attendance_courses$Event.Date.and.Time_2, "%H"))

attendance_courses$Date <-
  format.Date(attendance_courses$Event.Date.and.Time_2, "%m/%d/%Y")

# convert 00 hour to NA
# attendance_courses[attendance_courses$hour==00, "hour"] <- NA

attendance_courses$hour <- as.numeric(attendance_courses$hour)

# Semester defined as spring (month 1-5), summer (month 6-7), fall (month 8-12)

attendance_courses$Semester[attendance_courses$month>0&
                            attendance_courses$month<6] <- "Spring"
attendance_courses$Semester[attendance_courses$month>5&
                            attendance_courses$month<8] <- "Spring"
attendance_courses$Semester[attendance_courses$month>7&
                            attendance_courses$month<=12] <- "Spring"



###########
# Join outreach and course datasets
###########

# using dplyr's bind_rows instead of cbind ensures all columns are kept, including those unique to each dataset

attendance_named <- dplyr::bind_rows(attendance_outreach,
                                    attendance_courses)
                                    

###########

## attendance
attendance_named <- left_join(attendance_named,
                                 topics_complete_variations_cleaned,
                                 by =  c("Name.of.Workshop" = "WorkshopNameVariations")) %>% 
  dplyr::filter(!is.na(id))





###########
# Write to file for use in qmds.
###########
# Attendance
write.csv(attendance_named,
          file = "raw_data/processed_attendance_named.csv",
          row.names = FALSE)


```

- Post-workshop surveys

- Deal with all these data (attendance data are in ok now)


```{r}
###########
# Registration data has to be imported per-room from LibCal.
###########
# Event registration 
Room339 <- read.csv("./raw_data/preworkshop/Room339_lc_events_20230105043223.csv")
RoomLL123 <- read.csv("./raw_data/preworkshop/LL123_lc_events_20230105043507.csv")
RoomGeneral <- read.csv("./raw_data/preworkshop/General_lc_events_20230105043258.csv")
RoomLearningLabClassroom <- read.csv("./raw_data/preworkshop/LearningLabClassroom_lc_events_20230105043409.csv")

# Merge
Registered <- rbind(Room339,
                    RoomLL123,
                    RoomGeneral,
                    RoomLearningLabClassroom
)




###########
# Post-workshop survey data are from a subset of learners who attended and also filled out a survey.  In a few cases we forget to administer the survey or there's not enough time, or individuals opt out even when physically present in the room.
###########

# Post-workshop surveys from libwizard only
## Used for both feedback AND marketing
FeedbackActuallyAttended <- read.csv("./raw_data/postworkshop/report.csv",
                                     na.strings=c("","NA"))


library(tidyr)
library(stringr)

FeedbackActuallyAttended2 <- FeedbackActuallyAttended %>%
  separate_longer_delim(This.workshop.was...,
                        delim = ", ...") %>%
  separate_wider_delim(cols = This.workshop.was...,
                       delim = ". - ",
                       names = c("This_workshop_was",
                                 "Ranking"),
                       too_few = "align_start")

FeedbackActuallyAttended2$This_workshop_was <-
  str_remove(FeedbackActuallyAttended2$This_workshop_was,
             pattern = fixed("..."))

postworkshopsurveys_wide <- FeedbackActuallyAttended2 %>%
  pivot_wider(names_from = This_workshop_was,
              values_from = Ranking) %>%
  dplyr::select(-`NA`)



#rows <- as.numeric(count(postworkshopsurvey))

#postworkshopsurvey <- na.omit(postworkshopsurvey)




###########
# Changes in categories recorded over time have resulted in the need for some automated clean up code, which allows us to look at older categories if we need to, but use the updated equivalents for this report.
###########

# Creating and splitting current levels for marketing
marketing_current_categories <- c("Announcement in Class by Instructor",
                                  "Announcement in Class by Librarian",
                                  "Non-Library Calendar or Newsletter",
                                  "Another Libraries Workshop",
                                  "Walk-in",
                                  "Digital Signage in Non-Library Building",
                                  "Digital Signage in a Library Building",
                                  "Social Media - Other",
                                  "Social Media - Instagram",
                                  "Social Media - Facebook",
                                  "Social Media - Twitter",
                                  "Personal recommendation from OU Librarian",
                                  "Class requirement",
                                  "Departmental/program requirement",
                                  "OU Libraries website",
                                  "Direct Email from OU Librarian to Departmental Listserv",
                                  "Direct Email from OU Libraries Workshop Listserv",
                                  "Direct Email from OU Libraries (Source Unknown)",
                                  "Flyer (Paper or Digital Unspecified)",
                                  "Flyer (Paper or Paper Signage)",
                                  "Supervisor (Forwarded email, word of mouth)",
                                  "Colleagues (Forwarded email, word of mouth)")




## Convert old categories to equivalent new
Registered$How.did.you.hear.about.this.workshop.[Registered$How.did.you.hear.about.this.workshop.=="Email from University Libraries"] <- "Direct Email from OU Libraries (Source Unknown)"
Registered$How.did.you.hear.about.this.workshop.[Registered$How.did.you.hear.about.this.workshop.=="Colleagues"] <- "Colleagues (Forwarded email, word of mouth)"
Registered$How.did.you.hear.about.this.workshop.[Registered$How.did.you.hear.about.this.workshop.=="Supervisor"] <- "Supervisor (Forwarded email, word of mouth)"
Registered$How.did.you.hear.about.this.workshop.[Registered$How.did.you.hear.about.this.workshop.=="Department/Program Requirement"] <- "Departmental/program requirement"
Registered$How.did.you.hear.about.this.workshop.[Registered$How.did.you.hear.about.this.workshop.=="Flyer"] <- "Flyer (Paper or Digital Unspecified)"
Registered$How.did.you.hear.about.this.workshop.[Registered$How.did.you.hear.about.this.workshop.=="Walk-in"] <- "Other (please describe)" #correcting that walk-in should not be a pre-registration option

Registered$ReducedCategories <- Registered$How.did.you.hear.about.this.workshop.
Registered$ReducedCategories[!(Registered$How.did.you.hear.about.this.workshop. %in% marketing_current_categories)] <- "Other (please describe)"
Registered$OtherDescriptions[!(Registered$How.did.you.hear.about.this.workshop. %in% marketing_current_categories)] <- Registered$If.other..please.describe.[!(Registered$How.did.you.hear.about.this.workshop. %in% marketing_current_categories)]

## Convert old categories to equivalent new in feedback/post-workshop surveyd ata too.
postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.[postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.=="Email from OU Libraries"] <- "Direct Email from OU Libraries (Source Unknown)"
postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.[postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.=="Colleagues"] <- "Colleagues (Forwarded email, word of mouth)"
postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.[postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.=="Supervisor"] <- "Supervisor (Forwarded email, word of mouth)"
postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.[postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.=="Flyer"] <- "Flyer (Paper or Digital Unspecified)"

postworkshopsurveys_wide$ReducedCategories <- postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.
postworkshopsurveys_wide$ReducedCategories[!(postworkshopsurveys_wide$How.did.you.hear.about.this.workshop. %in% marketing_current_categories)] <- "Other (please describe)"
postworkshopsurveys_wide$OtherDescriptions[!(postworkshopsurveys_wide$How.did.you.hear.about.this.workshop. %in% marketing_current_categories)] <- postworkshopsurveys_wide$How.did.you.hear.about.this.workshop.[!(postworkshopsurveys_wide$How.did.you.hear.about.this.workshop. %in% marketing_current_categories)]


                                 
## registrations
registrations_named <- left_join(Registered,
                                 topics_complete_variations_cleaned,
                                 by =  c("Title" = "WorkshopNameVariations"))


## postworkshop surveys
postworkshopsurveys_named <- left_join(postworkshopsurveys_wide,
                                       topics_complete_variations_cleaned,
                                       by =  c("What.workshop.did.you.attend." = "WorkshopNameVariations"))






###########
# Month, day, year, hour, minute, seconds
###########
postworkshopsurveys_named$Date_2 <- ymd_hms(postworkshopsurveys_named$What.date.was.the.workshop.,
                                                           truncated = 3)


# separate year, month, date, times
postworkshopsurveys_named$year <- year(postworkshopsurveys_named$Date_2)
postworkshopsurveys_named$month <- month(postworkshopsurveys_named$Date_2)
postworkshopsurveys_named$day <- day(postworkshopsurveys_named$Date_2)
postworkshopsurveys_named$wday <- wday(postworkshopsurveys_named$Date_2,
                                      week_start = 1, #1 = Monday,
                                      label = TRUE    # days of weeks as characters
                                      )

# Overwrite original column with formatted date
postworkshopsurveys_named$Date <-
  format.Date(postworkshopsurveys_named$Date_2, "%m/%d/%Y")


###########
# Ordered scales require using R levels to order categories for later graphing or analysis.
###########

#This will need to be repeated for all three "valuable for" questions, so the levels are ordered in the plots.

# First, rename difficult columns
# rename(df, newname = oldname)
postworkshopsurveys_named <- postworkshopsurveys_named %>% 
  dplyr::rename(Valuable.for.Program = `valuable towards your program of study`, 
                Valuable.for.Teaching  = `valuable towards your teaching`,
                Valuable.for.Career = `valuable towards your career`,
                Using.New.Knowledge. = `Do.you.anticipate.that.this.new.knowledge.can.b...`)

## Next, order the levels for all three.
postworkshopsurveys_named$Valuable.for.Program <- factor(x = postworkshopsurveys_named$Valuable.for.Program,
                                                  levels = c("No answer",
                                                             "Not applicable", 
                                                             "Strongly agree",
                                                             "Agree",
                                                             "Somewhat agree",
                                                             "Neither agree nor disagree",
                                                             "Somewhat disagree",
                                                             "Disagree",
                                                             "Strongly disagree"
                                                             
                                                  ))


###########
# Finally, the tidied feedback and attendance data are combined to compare feedback with per-workshop characteristics such as format of instruction.  This is a smaller dataset as not all the workshops appear to have been entered into LibInsights, where the attendance and format data are stored.
###########

joined_post <- full_join(postworkshopsurveys_named, 
                         attendance_named,
                         by = c("Date", "WorkshopCode"))
# It is necessary to join by both date and workshop code as occasionally 2 or more people will teach on the same day in different events or classes.
# needs to be a full join because there are some courses not entered and vice versa.

# Filtering by NA is necessary because not everyone has entered their workshops into LibInsights and past workshops did not use the same feedback form, resulting in no attendance or format data to associate with the feedback or vice versa.  filtering by workshop code removes non-workshop course visits.
joined_attendance_post_clean <- joined_post %>%
  dplyr::filter(!is.na(WorkshopCode))

# Later data analysis will remove observations with no feedback automatically.


###########
# Write to files for use in qmds.
###########


# Satisfaction
write.csv(joined_attendance_post_clean,
          file = "raw_data/processed_joined_attendance_post_clean.csv",
          row.names = FALSE)

# Marketing

#not sure why this one is different
write.csv(postworkshopsurveys_named,
          file = "raw_data/processed_postworkshopsurveys_named.csv",
          row.names = FALSE)

write.csv(Registered,
          file = "raw_data/processed_Registered.csv",
          row.names = FALSE)

```