R/fars_functions.R

In alex2905/coursera: Test creation of a package

# Documenting functions
# Author: Alexander Kurtz
# Date: 2018-01-05
# Last update: 2018-01-05

#' Reading a csv file into R
#'
#' The function \code{fars_read} gets as input the path of the file name to load
#' into R. It checks whether the file exists and if not, stops.
#' If the file exists, with correct format, it is loaded into R with the
#' \code{read_csv} function from the \code{readr} package and messages,which
#' type each column is, are suppressed.
#' At least, the function uses the \code{tbl_df} function from the \code{dplyr}
#' package, although there is no need to do that.
#'
#' Erros occur by files which exist but are not in (zipped) csv format, or files
#' which are in csv format but do not have the default delimiters.
#' Error occurs if more than one filename is passed.
#'
#' @param filename A string containing the path for the file
#' @importFrom readr read_csv
#' @importFrom dplyr tbl_df
#' @return A tibble dataframe containing the information of the csv file
#' @export
#' @examples
#' # filename <- make_filename(2013)
#' # accident_2013 <- fars_read(filename)
fars_read <- function(filename) {
  if(!file.exists(filename))
    stop("file '", filename, "' does not exist")
  data <- suppressMessages({
    readr::read_csv(filename, progress = FALSE)
  })
  dplyr::tbl_df(data)
}


#' Creating a compromized file name
#'
#' The function \code{make_filename} creates a pre-defined name of a bz2 zipped
#' file, which is accident_year.csv.bz2. The year is the input of the function.
#' A warning occurs, if the input parameter is a single value but cannot be
#' converted into a integer, because year is then NA.
#'
#' An error occurs, if the input cannot be coerced to type integer - e.g. for
#' dataframes.
#'
#' @param year A string or number indicating the year for naming the file
#' @return A character string, indicating the zipped file name like
#'         accident_year.csv.bz2
#' @export
#' @examples
#' make_filename(2013)
make_filename <- function(year) {
  year <- as.integer(year)
  system.file("extdata", sprintf("accident_%d.csv.bz2", year),
              package = "coursera")
}


#' Reading the bz2 zipped files with corresponding year in their name and
#' selecting only months and years of this dataset.
#'
#' The function \code{fars_read_years} can load several files corresponding to
#' their year into R, since it gets a list/vector of years as input and returns
#' the month and years of that dataframes.
#' It is still possible to only load one dataset and manipulate it.
#' The lapply functionality allows to create for each element (year of years)
#' the zipped file name, check whether it exists with \code{tryCatch} and if it
#' throws an error, write out a warning that the year is invalid and return
#' \code{NULL}.
#' If the file exists, the inner of \code{tryCatch} is executed and the data is
#' loaded into R. Then a new column year is created, which is the current
#' element of years chosen in lapply and only Month and year of the dataset are
#' selected.
#'
#' Errors occur in the same cases as for \code{make_filename}, but not for
#' \code{fars_read}, since this is part of the tryCatch.
#'
#' @importFrom dplyr mutate select
#' @param years A list/vector containing the years for which data should be
#'              loaded and month + year are extracted
#' @return A dataframe containing the month and the year of each observation
#'         of the zipped file - the year coming as input and has to be part of
#'         the file name.
#' @export
#' @examples
#' # fars_read_years(2013)
fars_read_years <- function(years) {
  lapply(years, function(year) {
    file <- make_filename(year)
    tryCatch({
      dat <- fars_read(file)
      dplyr::mutate(dat, year = year) %>%
        dplyr::select(MONTH, year)
    }, error = function(e) {
      warning("invalid year: ", year)
      return(NULL)
    })
  })
}


#' Reading bz2 zipped data files and creating a summary of the number of
#' observations per month per year (years correspond to datasets)
#'
#' The function \code{fars_summarize_years} uses the \code{fars_read_years}
#' function toload the datasets and creates a list of each dataset containing
#' only the month and year per observation.
#' All this data is combined, via binding all rows together.
#' Next, the data is grouped by month and year that the number of observations
#' of each grouping can be measured.
#' In the last step, year and the number of observations per month and year
#' \code{(=n)} are spread, meaning new columns are created for each element of
#' the year column and it's value is the number of \code{n} in this month.
#'
#' Error occur, in the same cases as for \code{fars_read_years}.
#' If there are no errors in this help function, everything works well.
#'
#' @importFrom dplyr bind_rows group_by summarize
#' @importFrom tidyr spread
#' @inheritParams fars_read_years
#' @return A dataframe containing the number of each month and the number of
#'         observations for each month in each of the input years
#' @export
#' @examples
#' # fars_summarize_years(2013)
fars_summarize_years <- function(years) {
  dat_list <- fars_read_years(years)
  dplyr::bind_rows(dat_list) %>%
    dplyr::group_by(year, MONTH) %>%
    dplyr::summarize(n = n()) %>%
    tidyr::spread(year, n)
}


#' Plotting the locations of accidents for one state in one year with its bounds
#'
#' The function \code{fars_map_state} reads the bz2 zipped file into R via the
#' functions \code{make_filename} and \code{fars_read}.
#' Converts the state.num to integers.
#' Checks whether the state number exists in the dataset - if not, it stops and
#' writes out the invalid state number. If the state number exists, the data is
#' filtered by this state.
#' If there are no observations for this state the message "no accidents to plot"
#' appears. Seems unnecessary, since there have to be observations due to the
#' previous check.
#' LONGITUD values which exceed 900 are corrected to NA.
#' The same logic is applied for LATITUDE values over 90.
#' Then, the locations of the accidents in that state are plotted.
#'
#' Errors occur if there are errors in \code{make_filename} or \code{fars_read}
#' or the state.num cannot be converted to integer.
#' Additionally errors occur if the observations are out of bounds of the state.
#'
#' @importFrom dplyr filter
#' @importFrom maps map
#' @importFrom graphics points
#' @param state.num A number, indicating the number of states for filtering
#' @inheritParams make_filename
#' @return A plot, showing the locations (longitud and latitude) of accidents in
#'         the chosen state and year
#' @export
#' @examples
#' # fars_map_state(state.num = 9, year = 2013)
fars_map_state <- function(state.num, year) {
  filename <- make_filename(year)
  data <- fars_read(filename)
  state.num <- as.integer(state.num)

  if(!(state.num %in% unique(data$STATE)))
    stop("invalid STATE number: ", state.num)
  data.sub <- dplyr::filter(data, STATE == state.num)
  if(nrow(data.sub) == 0L) {
    message("no accidents to plot")
    return(invisible(NULL))
  }
  is.na(data.sub$LONGITUD) <- data.sub$LONGITUD > 900
  is.na(data.sub$LATITUDE) <- data.sub$LATITUDE > 90
  with(data.sub, {
    maps::map("state", ylim = range(LATITUDE, na.rm = TRUE),
              xlim = range(LONGITUD, na.rm = TRUE))
    graphics::points(LONGITUD, LATITUDE, pch = 46)
  })
}