Error in colmeans x na rm true x must be numeric

One error message you may encounter when using R is:

Error in colMeans(x, na.rm = TRUE) : 'x' must be numeric

This error usually occurs when you attempt to use the prcomp() function to perform principal components analysis in R, yet one or more of the columns in the data frame you’re using is not numeric.

There are two ways to get around this error:

Method 1: Convert Non-Numeric Columns to Numeric

Method 2: Remove Non-Numeric Columns from Data Frame

The following examples show how to use each method in practice.

How to Reproduce the Error

Suppose we attempt to perform principal components analysis on the following data frame that contains a character column:

#create data frame
df <- data.frame(team=c('A', 'A', 'C', 'B', 'C', 'B', 'B', 'C', 'A'),
                 points=c(12, 8, 26, 25, 38, 30, 24, 24, 15),
                 rebounds=c(10, 4, 5, 5, 4, 3, 8, 18, 22))

#view data frame
df

  team points rebounds
1    A     12       10
2    A      8        4
3    C     26        5
4    B     25        5
5    C     38        4
6    B     30        3
7    B     24        8
8    C     24       18
9    A     15       22

#attempt to calculate principal components
prcomp(df)

Error in colMeans(x, na.rm = TRUE) : 'x' must be numeric

The team column is a character column, which causes an error when we attempt to use the prcomp() function.

Method 1: Convert Non-Numeric Columns to Numeric

One way to avoid the error is to convert the team column to a numeric column before using the prcomp() function:

#convert character column to numeric
df$team <- as.numeric(as.factor(df$team))

#view updated data frame
df

  team points rebounds
1    1     12       10
2    1      8        4
3    3     26        5
4    2     25        5
5    3     38        4
6    2     30        3
7    2     24        8
8    3     24       18
9    1     15       22

#calculate principal components
prcomp(df)

Standard deviations (1, .., p=3):
[1] 9.8252704 6.0990235 0.4880538

Rotation (n x k) = (3 x 3):
                 PC1        PC2         PC3
team     -0.06810285 0.04199272  0.99679417
points   -0.91850806 0.38741460 -0.07907512
rebounds  0.38949319 0.92094872 -0.01218661

This time we don’t receive any error because each column in the data frame is numeric.

Method 2: Remove Non-Numeric Columns from Data Frame

Another way to avoid the error is to simply remove any non-numeric columns from the data frame before using the prcomp() function:

#remove non-numeric columns from data frame
df_new <- df[ , unlist(lapply(df, is.numeric))]

#view new data frame
df_new

  points rebounds
1     12       10
2      8        4
3     26        5
4     25        5
5     38        4
6     30        3
7     24        8
8     24       18
9     15       22

#calculate principal components
prcomp(df_new)

Standard deviations (1, .., p=2):
[1] 9.802541 6.093638

Rotation (n x k) = (2 x 2):
                PC1       PC2
points    0.9199431 0.3920519
rebounds -0.3920519 0.9199431

Once again, we we don’t receive any error because each column in the data frame is numeric.

Note: In most cases, the first method is the preferred solution because it allows you to use all of the data rather than removing some of the columns.

Additional Resources

The following tutorials explain how to fix other common errors in R:

How to Fix in R: Arguments imply differing number of rows
How to Fix in R: error in select unused arguments
How to Fix in R: replacement has length zero

When you are doing a Principal Components Analysis, you will get the “error in colmeans(x, na.rm = true) : ‘x’ must be numeric” error message if one of your columns has characters or other non-numeric values. Fortunately, there is a simple solution for fixing this problem. It simply involves translating a factor variable into a numeric variable.

Description of the error

This error message occurs because when you are doing a Principal Components Analysis, the values of each column of your data frame have to have numeric values. If it has characters or other non-numeric values such as missing values, you will get our error message. This occurs because the prcomp function only works with numeric values, so as a result, you will get an error message if the values are not numeric. As a result, if you need to run this kind of analysis, you need to make sure that you are giving it only numeric values. If you give it non-numeric values, you will get our error message.

Explanation of the error

The following example contains code that produces our error message. You should note column Z of the data frame.

> t = as.numeric(Sys.time())
> set.seed(t)
> z = c(“A”, “B”, “C”, “D”, “E”)
> x = rnorm(5)
> y = rnorm(5)
> df = data.frame(z, x, y)
> df
z x y
1 A 0.02307778 0.41365815
2 B 0.63213959 0.77502100
3 C -0.91366753 1.83374930
4 D 0.90422176 -0.09915274
5 E 0.75987927 -0.77146351
> pr = prcomp(df)
Error in colMeans(x, na.rm = TRUE) : ‘x’ must be numeric

If you look at data frame df you will notice that column Z has characters instead of numbers. It is this fact that triggers our error message because it is looking only for numeric values.
<h2>How to fix the error.</h2>Here we have an example of how to fix this problem. As long as you can convert the column into a factor, you can easily convert it into a numeric value. This is what we do in this example, and it fixes the problem.

> t = as.numeric(Sys.time())
> set.seed(t)
> z = c(“A”, “B”, “C”, “D”, “E”)
> x = rnorm(5)
> y = rnorm(5)
> df = data.frame(z, x, y)
> df
z x y
1 A 1.0158299 1.3621230
2 B -1.0393691 -0.4218296
3 C 0.1113177 0.5536360
4 D 1.8122020 1.1435097
5 E -1.3957393 0.9001602
> df2 = df
> df2$z = as.numeric(as.factor(df2$z))
> df2
z x y
1 1 1.0158299 1.3621230
2 2 -1.0393691 -0.4218296
3 3 0.1113177 0.5536360
4 4 1.8122020 1.1435097
5 5 -1.3957393 0.9001602
> pr = prcomp(df2)
> pr
Standard deviations (1, .., p=3):
[1] 1.664002 1.351141 0.469779

Rotation (n x k) = (3 x 3):
PC1 PC2 PC3
z 0.86675116 -0.4768597 -0.1461071
x -0.49435044 -0.7826437 -0.3782676
y -0.06603079 -0.4000920 0.9140932

If you will take a look at the difference between data frames df and df2, you will see that in df2 column Z is a series of numbers rather than letters. This conversion was accomplished by converting the column into a factor and then converting the factor into a list of numeric values.

This error message results from a simple mistake to make, but one that is also easy to fix. It is a simple matter of making sure that what you are putting through a Principal Components Analysis is only a numeric variable. This one simple correction will allow you to do the analysis without any errors. This means that you will get the results that you are looking for.