R/analysis.R

In wush978/DSRTeacher: Export Tracking Data of Students from DataScienceAndR

#'@title Source of Courses
#'@export
#'@description A named list. The name is the course name in the records. The value is another list,
#'where each list is the object returned from applying \code{yaml::yaml.load_file} to the course sources.
NULL

#'@export
DSR.lessons <- sprintf(
  "DataScienceAndR:%s",
  c("00-Hello-DataScienceAndR", "01-RBasic-01-Introduction", "01-RBasic-02-Data-Structure-Vectors",
    "01-RBasic-03-Data-Structure-Object", "01-RBasic-04-Factors",
    "01-RBasic-05-Arrays-Matrices", "01-RBasic-06-List-DataFrame",
    "01-RBasic-07-Loading-Dataset", "02-RDataEngineer-01-Parsing",
    "02-RDataEngineer-02-XML", "02-RDataEngineer-03-JSON", "02-RDataEngineer-04-Database",
    "02-RDataEngineer-05-Data-Manipulation", "02-RDataEngineer-06-Join",
    "03-RVisualization-01-One-Variable-Visualization", "03-RVisualization-02-Multiple-Variables-Visualization",
    "03-RVisualization-03-ggplot2", "03-RVisualization-04-Javascript-And-Maps",
    "Optional-RProgramming-01-Loop-And-Condition", "Optional-RProgramming-02-Function-Introduction",
    "Project-ROpenData-Power-GDP"
    )
)
#'@export
#'@title Analyze whether the student skipped the questions or not
#'@return data.frame with following columns: \itemize{
#'  \item course The course name.
#'  \item question.no The number of the question
#'  \item user_id The id of the student
#'  \item skipped Whether the question is skipped or not
#'  \item source.position integer. The index of which part of source found in \code{course.source}
#'}
#'Note that the students might do the question multiple times, so
#'there might be multiple rows for the specific course, question.no and user_id.
#'Each row represents each trials of the student at the specific course and question.
#'@param records The data.frame of the raw records. This should be the output of the function
#'\code{get.records}.
#'@param students The user_id to be analyzed. If it is \code{NULL}, then all students in the \code{records} will
#'be analyzed.
#'@param lessons The character vector of the courses to be analyzed.
#'@param course.source named list. The name is the course name.
#'The value is the root of the swirl course directory.
#'
#'@examples
#'\dontrun{
#'login("wush")
#'records <- get.records()
#'skipped.result <- skipped.analysis(records)
#'library(dplyr)
#'group_by(skipped.result, course, question.no, user_id) %>%
#'  summarise(is.skipped = all(skipped)) %>% # if the student does not skip once, then I recognize that he/she did the question.
#'  group_by(course, question.no) %>%
#'  summarise(skip.ratio = mean(is.skipped)) %>%
#'  arrange(desc(skip.ratio))
#'}
skipped.analysis <- function(
  records, students = NULL,
  lessons = DSR.lessons,
  course.source = course.sources
  ) {
  if (is.null(students)) students <- unique(records$user_id)
  as.data.frame(records) %>%
    dplyr::filter(user_id %in% students, course %in% lessons, type == 1) %>%
    dplyr::group_by(course) %>%
    dplyr::do({
      course.tokens <- .$course[1] %>%
        strsplit(split = ":", fixed = TRUE) %>%
        extract2(1)
      course <- course.tokens[1]
      lesson <- course.tokens[2]
      lesson.srcs <- course.source[[sprintf("%s:%s", course, lesson)]]
      lesson.classes <- lapply(lesson.srcs, lapply, "[[", "Class")
      dplyr::group_by(., user_id) %>%
        dplyr::do({
          user_id.data <- .
          # browser()
          log <- lapply(.$log, jsonlite::fromJSON)
          itertools::izip(
            correct = lapply(log, "[[", "correct"),
            skipped = lapply(log, "[[", "skipped"),
            question_number = lapply(log, "[[", "question_number")
          ) %>%
            lapply(function(x) {
              if (length(x$correct) > length(x$skipped)) x$skipped <- c(FALSE, x$skipped)
              result <- x$skipped[x$correct]
              .i <- which(
                sapply(lesson.classes, function(lesson.class) {
                  all(!lesson.class[x$question_number + 1] %in% c("meta", "text"))
                }) &
                  sapply(lesson.classes, function(lesson.class) {
                    sum(x$correct) == sum(!lesson.class %in% c("meta", "text"))
                  })
                )
              lesson.src <- lesson.srcs[[.i]]
              lesson.class <- lesson.classes[[.i]]
              data.frame(question.no = which(!lesson.class %in% c("meta", "text")), skipped = result, source.position = .i)
            }) %>%
            do.call(what = rbind)
        })
    }) %>%
    as.data.frame()
}