CODEWITHSUNDEEP
rbinaryhamming-distance
silverlight
silverlight

Reputation: 45

How do I calculate the hamming distance in R on a set of binary values?

I need to compute the hamming distance and plot them in clusters in R for a dataset that has 2 columns and 45,000+ rows. Is there well known library available for this? Or do any strategies come recommended stronger than others?

I tried the hamming.distance function from the package "e1071", and get the error below. But even if I figure out how to calculate the hamming distance, I am not sure how to transition from those results to a cluster graph? 

Error: evaluation nested too deeply: infinite recursion/options(expressions=)?
2015-02-02 18:50:59.704 R[1162:679616] Communications error:    <OS_xpc_error<error: 0x7fff7aaadb60> { count = 1, contents =
"XPCErrorDescription" => <string: 0x7fff7aaadfa8> { length = 22, contents =    "Connection interrupted" }

I tried this code:

 H<-hamming.distance(df)

Where df looks like this:

Name   Code
name1   0
name2   0
name3   1
name4   1
name5   0

Thank you for looking at this question and any help is greatly appreciated.

Upvotes: 1

Views: 2561

Answers (2)

Devon
Devon

Reputation: 690

To compare each row value to the previous row value, create a new column that is the previous row and apply this function across both columns.

df = data.frame(x1=as.character(c("0", "0", "1")))
df$x2 = c(NA, df$x1[-1])

hamming.distance = function(string1, string2){
  if (is.na(string2)==T) { 
    return (NULL)
  }
  string1 = as.character(string1)
  string2 = as.character(string2)

  length.string1 = nchar(string1)
  length.string2 = nchar(string2)

  if (length.string1 != length.string2) warning("Inputs must be of equal length")

 string.temp1 = c()
 for (i in 1:length.string1){
    string.temp1[i] = substr(string1, start=i, stop=i)
   }
  string.temp2 = c()
  for (i in 1:length.string2){
    string.temp2[i] = substr(string2, start=i, stop=i)
  }
   return(sum(string.temp1 != string.temp2))
}

results = mapply(hamming.distance, df[,1], df[,2])

unlist(results)

Note: the length of unlist(results) will be 1 shorter than the number of rows in your df object because the first entry is NA and unlist removes that value.

Upvotes: 2

Fedorenko Kristina
Fedorenko Kristina

Reputation: 2777

You can use stringdist package to calculate hamming distance: http://cran.r-project.org/web/packages/stringdist/stringdist.pdf

For example:

library(stringdist)
df <- data.frame( column1 = c("toned", "10112"), column2 = c("roses", "10223"))
stringdistmatrix(df$column1, df$column2, method = c("hamming"))#for distance matrix
stringdist(df$column1, df$column2, method = c("hamming"))#for vector of distance

Upvotes: 1

Related Questions