vignettes/model_details.md

In sherryxiu0305/SherryXiuCoursera: Accident Data Analysis

title: "Model details for SherryXiuCoursera Package" author: "Sherry Xiu" date: "2017-10-15" output: rmarkdown::html_vignette vignette: > %\VignetteIndexEntry{Model details for SherryXiuCoursera Package} %\VignetteEngine{knitr::rmarkdown} %\VignetteEncoding{UTF-8}

This package is designed to read csv files about accident data. Overall, the package has 2 main functions:

• fars_summarize_years() gives a 12-by-n table as the output, which summarizes the accident data of specific years into months
• fars_map_state() gives a map with marks as the output, which graphs the accident data of specified year and state onto the state map

Functions

There are a total of 5 functions in this package and they are all written in the same R file called far_functions.R. Below you can find a brief description of each function:

fars_read(filename)

This is a simple function that takes the file name string as parameter. If there exists such file, the function will read the file readr::read_csv and save the data in the file as a dataframe with dplyr::tbl_df. If the file name does not exist, the function will return an error stating that such file does not exist with stop().

Parameters

• filename: the filename (under current working directory) of the file interested

make_filename(year)

This is a function that takes the year as input. After turning it from a string into an integer with as.integer(), it prints out a file name in the form of "accident_n.csv.bz2",where n will be the year the user puts in. Notice that for this function, you could either input a string or an array of numbers and it will give you an array of file names made as well.

Parameters

• year: A single year of which the data that is interested.

fars_read_years(years)

This is a function that takes a vector of years as input. For each element in the vector, first, it makes the file name with the given year (using the make_filename() function mentioned above). Then it tries to read the data using fars_read() function and dat argument. For the data in the file, the function will select the 'month' variable of the elements where variable 'year' is equal to the current element in the years vector (with the tryCatch() function). If this process fails at any stage, it will return an error stating that the year is invalid. This process is to repeated for every element in the input vector.

Parameters

• years: An array of or a single year of which the data that is interested.

fars_summarize_years(years)

This function will take a vector of years as input. First, it reads through the files of given years and save the months and years data into a list using fars_read_years() function mentioned above. Then it turns the large list into a dataframe (with dplyr::bind_rows() function). After that, it groups the dataframe by month and year (using dplyr::group_by() function). Then, each group is summarized as the number of elements for each specific month in each year (using dplyr::summarize() function. In the end, to tidy up the output, this summary is displayed such that month is the element name and years are the variables spreading out (using tidyr::spread() function). The final result gives a 12-by-n table as the output, which summarizes the accident data of specific years into months

Parameters

• years: An array of or a single year of which the data that is interested.

fars_map_state(state.num,year)

This function turns accident data of specified year and state into map. It asks for state number and year from the user as inputs. It first makes the file name (using make_filename() function) and reads in the data from the file name (using fars_read() function). Then it ensure that the input state number is an integer (with as.integer() function). Then the function checks if the input state number is in the data. If it is in the data, the function will proceed to filter the data where only the observations that match for the given state number are selected. After that, the function will check if there are accidents happened in that states. If there are actually accidents, the function will start to map the accidents. First, it rules out any outlier in the selected data (with is.na() function). After all the cleaning and summarizing data, fhe function finally maps out the accidents as points representing their position in the states (using maps::map() graphics::points() functions).

Parameters

• state.num: the state number recorded in in the data given. If no match appears,the function will return an error

• year: the year of data wanted.

Example

Now a simple example is demostrated with a file named "accident_2015.csv.bz2".

First, we can implement the fars_summarize_years() function:

fars_summarize_years(2015)

## # A tibble: 12 x 2
##    MONTH `2015`
##  * <int>  <int>
##  1     1   2368
##  2     2   1968
##  3     3   2385
##  4     4   2430
##  5     5   2847
##  6     6   2765
##  7     7   2998
##  8     8   3016
##  9     9   2865
## 10    10   3019
## 11    11   2724
## 12    12   2781

In this function, '2015' is first passed into make_filename() to make a file name "accident_2015.csv.bz2". Then fars_read() would take the file name and read the data from that file. After that, fars_read_years() extracted the data about year and months from the data. These 3 functions are implemented inside of fars_summarize_years() so we don't need to call them separately. In the end, the filtered data is summarized into the output we see above.

Now we will show another example of using fars_map_state():

fars_map_state(1,2015)

In this function, "2015" is passed into make_filename() to make a file name so that fars_read() could read the data of the file, just like the previous function. After that, the function will extract all the data with the specified state number can plot the graph on the state map, as shown above.

sherryxiu0305/SherryXiuCoursera documentation built on May 29, 2019, 9:35 a.m.