How to convert factors into numeric

Question

My data:

data <- read.table(text = "Component    Method  Average StdErr
N170    Image   0,58904 0,0066948
P300    Image   0,84932 0,0048678
Sham    Image   0,23944 0,0058078
N170    Topo    0,64384 0,0065159
P300    Topo    0,90278 0,0040316
Sham    Topo    0,32877 0,0063921
N170    Joint   0,65278 0,0064788
P300    Joint   0,84722 0,0048959
Sham    Joint   0,26230 0,0059937", header = TRUE)

I would like to plot this using ggplot2. When trying to calculate the error bars, I get an error. I figured out that this is because "Average" is a factor and not a numeric here. Running my code:

ggplot(data=data, aes(x=Method, y=Average, group=Component)) + 
  geom_line(aes(colour=Component), size=1)+
  geom_pointrange(aes(ymin=Average-StdErr, ymax=Average+StdErr))+
  geom_point(aes(colour=Component), size=3)+
  theme_classic()

Gives me this error message:

Warning messages:
1: In Ops.factor(Average, StdErr) : ‘-’ not meaningful for factors
2: In Ops.factor(Average, StdErr) : ‘+’ not meaningful for factors
3: Removed 9 rows containing missing values (geom_pointrange).

I have tried converting Average and StdErr to numeric using this code, but it didnt solve my problem.

data$Average <- as.numeric(data$Average)
data$StdErr <- as.numeric(as.vector(data$StdErr))

Alternatively I tried reading in my data as a .csv:

data <- read.csv(file="C:/Users/XXX/Desktop/XXX/XX/ggplot/corrComp_agg.csv", 
                 header=TRUE, sep=",")

But get an error message saying:

Error in read.table(file = file, header = header, sep = sep, quote = quote,  : 
  more columns than column names

What am I missing?

@Moody, I think the problem isn't how to convert factors to numeric: I suspect the commas in the OP's numeric columns may be decimal places (i.e. originating from a locale that uses comma for decimal markers). If so adding `dec = ","` to the `read.table()` call will allow those columns to be read as character vectors, solving the problem more directly (and importing the data as numeric as intended). — krads, Sep 10 '18 at 11:07
relevant: https://stackoverflow.com/questions/3418128/how-to-convert-a-factor-to-integer-numeric-without-loss-of-information — moodymudskipper, Sep 10 '18 at 11:09
Very nice first question on Stack Overflow. Too many first-time posters on the R tag ask questions without providing for a way for others to reproduce the problem. I like how you began with a reproducible example. @krads is almost certainly right. You should find a way to read the data as numerical rather than fixing a bad read after the fact. — John Coleman, Sep 10 '18 at 11:19

score 2 · Answer 1 · answered Sep 10 '18 at 11:19

you should read the comma as decimal character.. using dec = "," in the read.table statement

data <- read.table(text = "Component    Method  Average StdErr
N170    Image   0,58904 0,0066948
                   P300    Image   0,84932 0,0048678
                   Sham    Image   0,23944 0,0058078
                   N170    Topo    0,64384 0,0065159
                   P300    Topo    0,90278 0,0040316
                   Sham    Topo    0,32877 0,0063921
                   N170    Joint   0,65278 0,0064788
                   P300    Joint   0,84722 0,0048959
                   Sham    Joint   0,26230 0,0059937", dec =",", header = TRUE)

Now your Average and StdErr are numeric

sapply(data, class)
#Component    Method   Average    StdErr 
# "factor"  "factor" "numeric" "numeric"

So, plotting is easy now

ggplot(data=data, aes(x=Method, y=Average, group=Component)) + 
  geom_line(aes(colour=Component), size=1)+
  geom_pointrange(aes(ymin=Average-StdErr, ymax=Average+StdErr))+
  geom_point(aes(colour=Component), size=3)+
  theme_classic()

How to convert factors into numeric

1 Answers1