How to Calculate Correlation Coefficients in R (5 Examples) | cor Function

This tutorial illustrates how to calculate correlations using the cor function in the R programming language.

The tutorial will consist of five examples for the application of the cor function. To be more specific, the content is structured as follows:

It’s time to dive into the examples!

Example Data

The data below is used as basement for this R programming tutorial.

First, we have to set a random seed for reproducibility:

set.seed(35843367)                          # Set random seed

Then, we can create a first variable as shown below:

x <- rnorm(100)                             # Create x variable
head(x)                                     # Print head of x variable
# [1]  0.5613421  0.3596981 -0.6503523  0.4343684  0.6023800  0.0320683

The previously shown output of the RStudio console shows that our example data is a randomly distributed numerical vector called x.

Next, we have to create a second variable:

y <- rnorm(100) + x                         # Create y variable
head(y)                                     # Print head of y variable
# [1]  0.5934054  2.0107541 -1.4445170 -1.2551753  1.5718713 -1.1317284

The previous output shows the first values of our second numerical variable.

Let’s use these data to calculate some correlations!

Example 1: Using cor() Function to Calculate Pearson Correlation

In this example, I’ll illustrate how to apply the cor function to compute the Pearson correlation coefficient.

Have a look at the following R code and its output:

cor(x, y)                                   # Pearson correlation
# [1] 0.63733

As you can see, the Pearson correlation coefficient of our two example variables is 0.63733.

Example 2: Using cor() Function to Calculate Kendall Correlation

We can also use the cor function to calculate other types of correlation coefficients. This example explains how to compute a Kendall Correlation:

cor(x, y, method = "kendall")               # Kendall correlation
# [1] 0.4719192

As shown in the previous R code, we had to set the method argument to be equal to “kendall”.

Example 3: Using cor() Function to Calculate Spearman Correlation

Similar to Example 2, we can use the method argument of the cor function to return the Spearman correlation coefficient for our two variables:

cor(x, y, method = "spearman")              # Spearman correlation
# [1] 0.6522172

Example 4: Calculate Correlation of Data with NA Values

In this example, I’ll explain how to calculate a correlation when the given data contains missing values (i.e. NA).

First, we have to modify our example data:

x_NA <- x                                   # Create variable with missing values
x_NA[c(1, 3, 5)] <- NA
head(x_NA)
# [1]        NA 0.3596981        NA 0.4343684        NA 0.0320683

As you can see in the RStudio console, we have inserted some NA values in our x variable.

If we now use the new x variable with NA values to calculate a correlation, NA is returned as result:

cor(x_NA, y)                                # Try to calculate correlation
# [1] NA

If we want to remove those NA observations from our data to calculate a valid correlation coefficient, we have to set the use argument to be equal to “complete.obs”:

cor(x_NA, y, use = "complete.obs")          # Remove NA from calculation
# [1] 0.6317544

Example 5: Calculate Correlation Matrix for Entire Data Frame

In Example 5, I’ll demonstrate how to create a correlation matrix for an entire data frame.

For this, we first have to create an exemplifying data set:

data <- data.frame(x, y, z = rnorm(100))    # Create example data frame
head(data)                                  # Print head of example data frame

Table 1 illustrates the first lines of our example data.

Next, we can use the cor function to calculate a correlation matrix of these data:

cor(data)                                   # Create correlation matrix

In Table 2 it is shown that we have created a correlation matrix for our example data frame by using the previous syntax.

Video & Further Resources

In case you need more information on the topics of this tutorial, I recommend having a look at the following video on the Statistics Globe YouTube channel. In the video, I illustrate the examples of this article:

In addition, you might want to read the other tutorials on this website:

• Correlation Matrix in R
• Correlation of One Variable to All Others
• Calculate Correlation Matrix Only for Numeric Columns
• Remove Highly Correlated Variables from Data Frame
• Variance in R
• Standard Deviation in R
• Useful Commands in R
• R Programming Language

To summarize: You have learned in this article how to compute correlations using the cor function in the R programming language. If you have any additional comments or questions, let me know in the comments.