Overview

In this assignment, information about my favorite books on one of my favorite subject were stored in three (3) different formats. These are HTML, XML, and JSON table formats. Each were created “by hand”, for which the codes are shown in the respective sections. The tables were saved to my Github and were retrieved from there.

The task was to create and practice loading these tables into R. No tidying or transformation was done, since I compared them based on the raw conversion from the specific format into an R data frame. Later, I discussed the advantages and disadvantages of each format I noticed while completing this assignment, and concluded on which format might be the better choice based on one’s purpose.

The main R packages that were need are below:

# Packages for working with HTML, XML & JSON
library(RCurl)
library(XML)
library(jsonlite)

# Other packages
library(kableExtra)

HTML format

All internet browsers support table formats. Visually, an HTML table is a simple layout of row and columns. Each cell can contain any class of data, from text, numerical, images, etc. HTML creates a neat and ordered depiction for its tables over a range of browsers due to its adaptability. Specifying the aesthetics of the table, such as shades, drop down capabilities, size, positioning, etc., further allows for more interactive and clearer presentation of information.

Let’s look at the HTML script and load it into R.

HTML Script

This snippet of HTML script creates a table of my favorite books.

Click to show HTML Script 

<html>
<head> <style>
table {
  font-family: arial, sans-serif;  border-collapse: collapse;  width: 100%;
}

td, th {
  border: 1px solid #dddddd;  text-align: left;  padding: 8px;
}

</style> </head> <body>

<h2>HTML Table of Favorite Books</h2>

<table id='FavoriteBooks'>
  <tr>
    <th>Book Title</th>
    <th>Authors</th>
    <th>Edition</th>
    <th>Year Published</th>
    <th>ISBN-10</th>
    <th>Publisher</th>
  </tr>
  <tr>
    <td>Handbook of Applied Multivariate Statistics and Mathematical Modeling</td>
    <td>Howard E.A. Tinsley, Steven D. Brown</td>
    <td>1st</td>
    <td>2000</td>
    <td>0126913609</td>
    <td>Academic Press</td>
  </tr>
    <tr>
    <td>Applied Multivariate Statistical Analysis</td>
    <td>Richard A. Johnson, Dean W. Wichern</td>
    <td>6th</td>
    <td>2007</td>
    <td>0131877151</td>
    <td>Pearson</td>
  </tr>
  <tr>
    <td>Applied Regression Analysis and Generalized Linear Models</td>
    <td>John Fox</td>
    <td>2nd</td>
    <td>2008</td>
    <td>0761930426</td>
    <td>Sage Publications</td>
  </tr>
</table>

</body>
</html>

Loading the File

Step 1: Use RCurl::getURL to download the URL for the raw data into R.

ptm <- proc.time()
htmlURL <- getURL("https://raw.githubusercontent.com/greeneyefirefly/Data607/master/HomeWork/HW5/books.htlm")

Step 2: Use XML::readHTMLTable to extract and parse the data from HTML tables in an HTML document. Convert the table into a data frame.

books.htlm <- as.data.frame(readHTMLTable(htmlURL, header = TRUE, fill = FALSE))
proc.time()-ptm
##    user  system elapsed 
##    0.49    0.26    0.80

Note that above is the time it took R to retrieve and convert the HTML file into a data frame. On average, it takes ~0.81 second for the code chunk to run. This made HTML file the longest to import.

The Output

Here is how the HTLM file looks like after it was parsed and converted into a data frame. It was noted that it retains the same structure as the HTML table. While it concatenate the main table title with the column names, it is still properly labeled. Moreover, each variable was stored as factor in the data frame.

FavoriteBooks.Book.Title FavoriteBooks.Authors FavoriteBooks.Edition FavoriteBooks.Year.Published FavoriteBooks.ISBN.10 FavoriteBooks.Publisher
Handbook of Applied Multivariate Statistics and Mathematical Modeling Howard E.A. Tinsley, Steven D. Brown 1st 2000 0126913609 Academic Press
Applied Multivariate Statistical Analysis Richard A. Johnson, Dean W. Wichern 6th 2007 0131877151 Pearson
Applied Regression Analysis and Generalized Linear Models John Fox 2nd 2008 0761930426 Sage Publications

XML format

XML is a markup language that uses human language to be read by machine. It can be used to share information in a consistent way as long as the XML document is considered to be “well formed” (i.e. easily parsed) and valid. That is, its format correctly with the XML specification, properly marked up, and nested properly.

Let’s look at the XML script and load it into R.

XML Script

This is a snippet of the XML script which creates a table of my favorite books.

Click to show XML Script 

<XML_Table_of_Favorite_books>
    <book_information>
        <Book_Title>Handbook of Applied Multivariate Statistics and Mathematical Modeling</Book_Title>
        <Authors>Howard E. A. Tinsley, Steven D. Brown</Authors>
        <Edition>1st</Edition>
        <Year_Published>2000</Year_Published>
        <ISBN-10>0126913609</ISBN-10>
        <Publisher>Academic Press</Publisher>
    </book_information>
    <book_information>
        <Book_Title>Applied Multivariate Statistical Analysis</Book_Title>
        <Authors>Richard A. Johnson, Dean W. Wichern</Authors>
        <Edition>6th</Edition>
        <Year_Published>2007</Year_Published>
        <ISBN-10>0131877151</ISBN-10>
        <Publisher>Pearson</Publisher>
    </book_information>
    <book_information>
        <Book_Title>Applied Regression Analysis and Generalized Linear Models</Book_Title>
        <Authors>John Fox</Authors>
        <Edition>2nd</Edition>
        <Year_Published>2008</Year_Published>
        <ISBN-10>0761930426</ISBN-10>
        <Publisher>Sage Publications</Publisher>
    </book_information>
</XML_Table_of_Favorite_books>

Loading the File

Step 1: Use RCurl::getURL to download the URL for the raw data into R.

ptm <- proc.time()
xmlURL <- getURL("https://raw.githubusercontent.com/greeneyefirefly/Data607/master/HomeWork/HW5/books.xml")

Step 2: Use XML::xmlParse to parse the data from XML file containing XML content and form an XML tree.

books.xml <- xmlParse(xmlURL)

Step 3: Use XML::xmlToDataFrame to convert the XML content into a data frame.

books.xml <- xmlToDataFrame(books.xml)
proc.time()-ptm
##    user  system elapsed 
##    0.08    0.04    0.19

On average, the time it takes R to retrieve and convert the XML file into a data frame was ~0.21 second for the code chunk to run. This made XML file the second longest to import.

The Output

Here is how the XML file looks like after it was converted into a data frame. It was noted that it also retains the same structure and still properly labeled as the script was programmed. Similar to the HTML data frame, each variable was stored as factor.

Book_Title Authors Edition Year_Published ISBN-10 Publisher
Handbook of Applied Multivariate Statistics and Mathematical Modeling Howard E. A. Tinsley, Steven D. Brown 1st 2000 0126913609 Academic Press
Applied Multivariate Statistical Analysis Richard A. Johnson, Dean W. Wichern 6th 2007 0131877151 Pearson
Applied Regression Analysis and Generalized Linear Models John Fox 2nd 2008 0761930426 Sage Publications

JSON format

When exchanging data between a browser and a server, the data can only be in text format, which can be interpreted as a JavaScript object. There is no complicated parsing, therefore making it is easy for humans to read and create, and easy for computers to generate.

Let’s look at the JSON script and load it into R.

JSON Script

This is a snippet of the JSON script which creates a table of my favorite books.

Click to show JSON Script 

{
    "JSON_Table_of_Favorite_books": [
            {
                "Book_Title": "Handbook of Applied Multivariate Statistics and Mathematical Modeling",
                "Authors": "Howard E. A. Tinsley, Steven D. Brown",
                "Edition": "1st",
                "Year_Published": "2000",
                "ISBN-10": "0126913609",
                "Publisher": "Academic Press"
            },
            {
                "Book_Title": "Applied Multivariate Statistical Analysis",
                "Authors": "Richard A. Johnson, Dean W. Wichern",
                "Edition": "6th",
                "Year_Published": "2007",
                "ISBN-10": "0131877151",
                "Publisher": "Pearson"
            },
            {
                "Book_Title": "Applied Regression Analysis and Generalized Linear Models",
                "Authors": "John Fox",
                "Edition": "2nd",
                "Year_Published": "2008",
                "ISBN-10": "0761930426",
                "Publisher": "Sage Publications"
            }
    ]
}

Loading the File

Step 1: Use RCurl::getURL to download the URL for the raw data into R

ptm <- proc.time()
jsonURL <- getURL("https://raw.githubusercontent.com/greeneyefirefly/Data607/master/HomeWork/HW5/books.json")

Step 2: Use jsonlite::fromJSON to parse the data from a JSON file into R objects. Then convert the JSON content into a data frame.

books.json <- as.data.frame(fromJSON(jsonURL))
proc.time()-ptm
##    user  system elapsed 
##    0.05    0.03    0.12

On average, it takes ~0.17 second for the code chunk to run that retrieve and convert the JSON file into a data frame. This made JSON file the quickest to import.

The Output

Here is how the JSON file looks like after it was parsed into a data frame. It retains the same structure and concatenate the main table title with the column names, similarly to how HTML did. However, in this case, each variable was stored as character.

JSON_Table_of_Favorite_books.Book_Title JSON_Table_of_Favorite_books.Authors JSON_Table_of_Favorite_books.Edition JSON_Table_of_Favorite_books.Year_Published JSON_Table_of_Favorite_books.ISBN.10 JSON_Table_of_Favorite_books.Publisher
Handbook of Applied Multivariate Statistics and Mathematical Modeling Howard E. A. Tinsley, Steven D. Brown 1st 2000 0126913609 Academic Press
Applied Multivariate Statistical Analysis Richard A. Johnson, Dean W. Wichern 6th 2007 0131877151 Pearson
Applied Regression Analysis and Generalized Linear Models John Fox 2nd 2008 0761930426 Sage Publications

Comparing the Formats

The Advantages Among the Formats

Some of the advantages I can already see with a beginner’s knowledge about these three methods are:

HTML
  • Every browser supports HTML language.
  • Outputs are clean and can be easily render.
  • It is widely used and well-known in spite of its disadvantages.
  • Can get very creative with font, colors, layouts, etc. for web-design.
  • Largely case insensitive, except for certain attributes, such as IDs.
XML
  • XML tag are comprehensible and easily convey the meaning of the data to both users and computers, similar to JSON.
  • Each XML tag immediately follows the respective data input, like JSON.
  • The data structure follows a noticeable pattern, making it easy to manipulate and exchange data, similar to JSON.
  • Loads faster into R than HTML.
JSON
  • Smaller and more compact script size.
  • JSON tag names are readable to both humans and computers, like XML.
  • Each JSON tag immediately follows the data, similar to XML.
  • The data structure follows a noticeable pattern, like XML
  • Can easily distinguish between the number and the string as numbers.
  • Takes less memory to store the same information.
  • Loads fastest into R than HTML and XML.

The Disadvantages Among the Formats

There are also some of disadvantages which are immediately noticeable about these three methods. These include:

HTML
  • HTML tag names reveal nothing about the meaning of the content.
  • HTML script is larger in size for a simple table.
  • Takes more memory to store the same information, similar to XML.
  • Loads slower into R than JSON and XML.
XML
  • Takes more memory to store the same information, similar to HTML.
  • XML does not contain any information indicating how the document should be rendered in a browser, unlike HTML.
  • Case sensitive and compulsory to use closing tags, unlike HTML.
JSON
  • Every single comma, quote, and bracket must be correctly place, unlike HTML and XML.
  • JSON does not contain any information indicating how the document should be rendered in a browser, unlike HTML.

Conclusion

Based on one’s purpose, if there is a need for web page design, HTML is the master, whereas XML can be a better format for document markup and JSON is better for structured data exchange.