Advanced Programming Concepts in R

In the first article, we covered the R programming language basics. In this article, we will look at some of the advanced aspects of the R programming language. Going through this article will enhance a developer’s ability to implement Object-Oriented Programming (OOP) concepts and write modular code. We will be covering the following topics in this article:

• Data Frames
• Typecasting
• File Handling
• Working with Inbuilt Datasets
• Conditional Statements
• Loop Statements

Data frames

R is a programming language built for statistical analysis of large datasets. Data frames help by storing large datasets in the local memory. They enable operations such as Create, Read, Update, and Delete (CRUD) to be performed efficiently. Formally, they are defined as: A table or a two-dimensional array-like structure in that each column contains values of one variable, and each row contains one set of values from each column”.

Let’s look at an example to define a data frame. We will describe several vectors storing different data types. The example considered below is a simplified version of a real dataset. Real datasets have much more variation with respect to the data.

For instance, we use the alphabet as dummy data. Real datasets have much more detailed information such as nouns, verbs, adjectives, etc. Real datasets contain columns of different data types. The data most often belong to the Boolean, character, or numeric data types.

vector_1 <- c("a","b","c","d","e") # creating a vector of 5 characters
vector_2 <- c("A","B","C","D","E") # creating a vector of 5 different characters
vector_3 <- c(1,2,3,4,5) # creating a vector with numerical entries
vector_4 <- c(2,4,6,8,10)
vector_5 <- c(TRUE,TRUE,TRUE,FALSE,FALSE) # creating a vector with boolean values

# creating a data frame out of 5 vectors.
new_data_frame <- data.frame(vector_1, vector_2, vector_3, vector_4, vector_5)
new_data_frame

We have defined a new data frame called new_data_frame. Data frames are tightly coupled collections of variables that share many of the matrices and lists’ properties. Therefore, defining data frames gives us access to most of the functions defined for matrices and lists. To define a new data frame, we use the data.frame method and pass the vectors as arguments.

The output of the above code is as follows:

    vector_1 vector_2 vector_3 vector_4 vector_5
1        a        A        1        2     TRUE
2        b        B        2        4     TRUE
3        c        C        3        6     TRUE
4        d        D        4        8    FALSE
5        e        E        5       10    FALSE

It is important to create data frames out of vectors to perform preprocessing on the datasets. There are primarily two types of analysis: Continuous and Categorical.

Continuous analysis deals with all continuous values (numbers) in the dataset, whereas categorical analysis deals with categories. We looked at the definitions of categorical and continuous features in the previous article under the factors section. For more information, check out the previous article.

Examples for the two are:

1. Statistical values: Prices of houses, oxygen levels in the blood, etc.
2. Categorical values: Names of places or objects, status (example: Sell, Rent, Buy)

R provides a few functions to help analyze data frames:

• head(): Prints out the first five rows of the data frame. This is used to preview the dataset and get information about the dimensionality (number of rows and columns) of the dataset.
• str(): Prints the structure of the data frame. The structure includes information about the data frame and the data types present in each vector. Notice that the data frame auto-assigns the data type to the vectors.
• tail(): Prints out the last six observations of the data frame.

The outputs for the functions head and tail are the same in this case. The two functions display the first five and last five observations, respectively. If we had more than five rows, the difference between the two functions could be observed.

    vector_1 vector_2 vector_3 vector_4 vector_5
1        a        A        1        2     TRUE
2        b        B        2        4     TRUE
3        c        C        3        6     TRUE
4        d        D        4        8    FALSE
5        e        E        5       10    FALSE

These functions make sense when the data frame size is enormous.

str function provides the following preview, with the data type and dataset for 5 observations.

data.frame':    5 obs. of  5 variables:
$ vector_1: Factor w/ 5 levels "a","b","c","d",..: 1 2 3 4 5
$ vector_2: Factor w/ 5 levels "A","B","C","D",..: 1 2 3 4 5
$ vector_3: num  1 2 3 4 5
$ vector_4: num  2 4 6 8 10
$ vector_5: logi  TRUE TRUE TRUE FALSE FALSE

Observe that vector_1 is assigned to a factor data type. To brush up on factor data type, revisit the previous article.

Data frames support indexing. The reasoning behind indexing is similar to that of the matrices. Let’s consider a few examples:

• Select the third column
• Select the first three rows
• Select the first three rows from the fifth column

The subsequent outputs is given below:

# Consider the same data_frame defined above: new_data_frame
new_data_frame[,3] # select third column
new_data_frame[1:3,] # select the first three rows
new_data_frame[1:3,5] # select the (1,5), (2,5) and (3,5)

The operations above output the following: The first output is the third column. The second part of the output is the first three rows. The last piece of the output denotes the three elements (1,5), (2,5) and (3,5).

We discussed the slicing and indexing operations in the previous article in depth. The example uses indexing and slicing operation. For a quick refresher on slicing, refer to the selection of elements subsection under the matrices section in the previous article.

    1 2 3 4 5

vector_1 vector_2 vector_3 vector_4 vector_5
1        a        A        1        2     TRUE
2        b        B        2        4     TRUE
3        c        C        3        6     TRUE

    TRUE TRUE TRUE

A final function to consider is subset. The subset function is similar to Python’s filter function. We can specify conditions using a subset and get a slice of the data frame, satisfying the condition.

Example: Let the condition be all those rows where vector_5 is not equal to FALSE.

subset(new_data_frame, subset= vector_5!=FALSE)

The output is given as follows. The subset function with parameter vector_5!=FALSE makes sure the condition is satisfied so that any row under vector_5 that is FALSE will not be displayed.

vector_1 vector_2 vector_3 vector_4 vector_5
1        a        A        1        2     TRUE
2        b        B        2        4     TRUE
3        c        C        3        6     TRUE

Typecasting

Similar to other programming languages like C++, Python, etc., R provides us with an option to explicitly change or modify the data type of previously defined variables. R offers the keyword as to typecast previously defined variables.

We will use the as keyword in the following example below:

x <- 0:6
class(x)
# output: integer
typecast_to_numeric <- as.numeric(x)
class(typecast_to_numeric) # integer -> numeric
# output: numeric
typecast_to_logical <- as.logical(x)
class(typecast_to_logical) # integer-> logical
# output: logical
typecast_to_char <- as.character(x)
class(typecast_to_char) # integer -> character
# output: character

Observe the change in the data types. An application of typecasting makes sense on user-defined data types. Consider the following applications:

• Consider a data frame with a column containing Boolean values. If one needs to convert these into integers to train a machine learning model, then typecasting is used. Typecasting converts the Boolean values into integer data.
• Suppose one is given a data frame with one column containing time in seconds. To convert each column of the provided data frame to time in days format, explicit typecasting is used.

For more information on explicit typecasting, refer to this link.

Reading and Writing Data

R provides many functions for reading and writing data. There are mainly three data formats used in R. They are .txt, .csv, and .RData.

The functions are defined for each of the file-types and are given as follows:

1. read.table: read.table("test.txt")
2. read.csv: read.csv("test.csv")
3. load: load("file_name.RData")

For writing into files, R provides the following functions:

1. write.table: write.table(data, file='file_name.txt')
2. write.csv: write.csv(data, file='file_name.csv')
3. save: save(x,y,"file_name.Rdata")

Whether reading from or writing to files, ensure that the file directory is correctly mentioned. This works when the file is in the same directory, otherwise, set up a relative link. Relative paths are used extensively in file handling.

Consider going through this answer on StackOverflow for a glance on relative paths.

Accessing Built-in Datasets

While learning various machine learning algorithms, the best place to start is the built-in datasets. These are small in size and are cleaned thoroughly. Clean datasets mean they have been pre-processed. The majority (80%) of a data scientist’s time is spent on data collection and data cleaning. Therefore, as beginners, we can skip this time-consuming step and deal with a tried and tested dataset.

These datasets were the benchmark datasets a decade ago. Today, beginners in data science use these as tools to learn and implement the various machine learning algorithms. In the description below, we have linked to a few sample projects using these datasets are given.

A few of the datasets that R provides are as follows:

• Airquality dataset: Dataset monitoring the air quality and the effect of air quality on weather and temperature. This dataset helps one understand the correlation between air quality and temperature. This article goes through exploring this dataset in full detail.

 head(airquality)

This dataset can be called through the variable airquality.

 Ozone Solar.R Wind Temp Month Day
 1    41     190  7.4   67     5   1
 2    36     118  8.0   72     5   2
 3    12     149 12.6   74     5   3
 4    18     313 11.5   62     5   4
 5    NA      NA 14.3   56     5   5
 6    28      NA 14.9   66     5   6

• Iris Dataset: The iris dataset provides four features for three species of flowers: Versonica, Setosa, and Virginica. The features are the sepal length, sepal width, petal length, and the petal width.

head(iris)

Outputs:

Sepal.Length Sepal.Width Petal.Length Petal.Width Species
1          5.1         3.5          1.4         0.2  setosa
2          4.9         3.0          1.4         0.2  setosa
3          4.7         3.2          1.3         0.2  setosa
4          4.6         3.1          1.5         0.2  setosa
5          5.0         3.6          1.4         0.2  setosa
6          5.4         3.9          1.7         0.4  setosa

For a thorough list, refer to the link mentioned here. Next, we will work with flow control and modularizing the code using conditionals and functions in R.

Conditional and Loop Statements

1. If-else statements: If-else statements allows one to implement sequential logic. These are an important aspect of OOP paradigm.

Let’s look at the example given below:

    x <- runif(1,0,10)
    if(x>3){
        y<-10
    }
    else{
        y <-0
    }

The runif function randomly picks n number of samples in the specified range [min, max). [ denotes closed interval, whereas ) denotes open interval. In the example given above, the runif function picks 1 sample randomly in the range [0,10) and assigns it to variable x. n is 1, min is 0, and max is 10.

The arguments to the function runif are:

• n: Number of samples
• min: lower bound of the range
• max: upper bound of the range

2. For loops: To demonstrate for loops, we will consider printing the elements of a list and a matrix. Since a matrix is a 2-D data structure, we will need two loops. Such usage of a loop within a loop is referred to as nested looping.

    for(i in 1:10){
        print(i)
    }
    # double for loop
    x<- matrix[1:6, 2:3]
    x
    for(i in seq_len(nrow(x))){
        for(j in seq_len(ncol(x))){
            print(x[i,j])
        }
    }

Notice the new seq_len function. We have just learned a new function. This is similar to the range function in Python. seq_len(x) generates integers from 1 to x. Note that R is a 1-indexed language. All indices start with one. We look at nested loops in the above example. The code selects the (i,j)th element from the matrix and prints it on the console. Recall that indices in R begin with 1.

The output for the code above is as follows:

        > x
            [,1] [,2] [,3]
        [1,]    1    3    5
        [2,]    2    4    6

        <!-- for loop output -->

        [1] 1
        [1] 3
        [1] 5
        [1] 2
        [1] 4
        [1] 6

3. While loops: While loops are known for their conditional execution of loops. They lack the iterator variable that the for loop uses. Let’s look at the syntax of while loops in R:

    while (condition_expression)
    {
    execute_code_while_condition_is_true
    }

Consider the following example:

    iter <-10
    while (iter>3){
        print(iter)
        iter<iter-1
    }

The output of the above is as follows. The code inside the loop executes until the value of iterator is less than or equal to 3. Hence, it prints the values till 4. When iter equals 3, it exits the loop.

    10
    9
    8
    7
    6
    5
    4

When the variable iter equals three, the condition fails. Therefore, the print statement stops executing.

Functions

Functions are important aspects of all programming languages. They help modularize the code. Modularization refers to the concept of breaking down a problem into the smallest pieces. This helps make debugging easier and provides better readability of code. Functions help by implementing various programming paradigms such as Don’t Repeat Yourself – DRY and modularity.

Let’s look at the following function definition in R:

function_name <- function(parameter_1=10){
    hello <- "Hellooooo" # assigns string to hello variable
    for i in seq_len(parameter_1){
        print(hello)
    }
    return (parameter_1 + 10)
}

To call the function, we pass the parameter to the function name directly:

function_name(10)
# output: prints hello 10 times in separate lines

In cases when multiple values need to be returned, we use lists. This way, a single entity with various outputs is being returned. I suggest you try this last point as a challenge.

Conclusion

In this article, we have dealt with the advanced topics of the R programming language. At this point, you have the skills to code machine learning algorithms from scratch and perform various data science experiments.