k nearest neighbor from Machine Learning for Hackers

knn.R

#
# read data frame from csv
#
df <- read.csv('data/example_data.csv')

#
# produce distance matrix
#
distance.matrix <- function(df)
  {
    distance <- matrix(rep(NA, nrow(df)^2), nrow = nrow(df))

    for(i in 1:nrow(df))
      {
        for(j in 1:nrow(df))
          {
            distance[i, j] <- sqrt( (df[i, 'X']-df[j, 'X'])^2 + (df[i, 'Y']-df[j, 'Y'])^2 )
          }
      }

    return(distance)
  }
                                   
#
# k nearest neighbors
#
k.nearest.neighbors <- function(i, distance, k = 5)
  {
    # sort i th row, return distance[i][2], distance[i][3], ..., distance[i][k+1]
    return(order(distance[i, ])[2:(k+1)])
  }

#
# knn for data frame
#
knn <- function(df, k = 5)
  {
    distance <- distance.matrix(df)

    predictions <- rep(NA, nrow(df))

    for(i in 1:nrow(df))
      {
        indices <- k.nearest.neighbors(i, distance, k = k)
        predictions[i] <- ifelse(mean(df[indices, 'Label']) > 0.5, 1, 0)
      }

    return(predictions)
  }

#
# append kNNPredictions
#
df <- transform(df, kNNPredictions = knn(df))