The k-nearest neighbors (k-NN) algorithm is a simple yet powerful tool used in various machine learning and data mining applications. While k-NN is often applied to an entire dataset to classify or predict values for multiple points, there are scenarios where you may need to find the k-nearest neighbors for a single point. This article provides a step-by-step guide to doing this in R Programming Language.

k Nearest Neighbor in R

k-NN is a non-parametric method used for classification and regression. The algorithm works by identifying the k closest points in the training set to a given point based on a distance metric, typically Euclidean distance. This article focuses on finding the k-nearest neighbors for a single point using R, a popular programming language for statistical computing and graphics.

Prerequisites

Before proceeding, ensure you have the following:

• Basic understanding of R programming.
• Installed R and RStudio (optional but recommended).
• The class and FNN packages installed. You can install them using install.packages(“class”) and install.packages(“FNN”).

Step 1. Load Necessary Libraries

First, load the required libraries:

library(class)
library(FNN)

Step 2. Prepare Your Data

For demonstration, we’ll create a sample dataset. Assume you have a dataset with two features and you want to find the k-nearest neighbors for a specific point.

# Sample dataset
set.seed(123)
data <- data.frame(
  x = rnorm(100, mean = 5, sd = 2),
  y = rnorm(100, mean = 5, sd = 2)
)

# Point of interest
point <- data.frame(x = 6, y = 6)

Step 3. Calculate k-Nearest Neighbors

Use the get.knnx function from the FNN package to find the k-nearest neighbors. This function returns the indices and distances of the k-nearest neighbors.

k <- 3
knn_result <- get.knnx(data, point, k)

# Indices of the nearest neighbors
indices <- knn_result$nn.index

# Distances to the nearest neighbors
distances <- knn_result$nn.dist

Step 4. Display the Results

Print the indices and distances of the k-nearest neighbors:

print("Indices of the nearest neighbors:")
print(indices)

print("Distances to the nearest neighbors:")
print(distances)

# Nearest neighbors' data points
nearest_neighbors <- data[indices, ]
print("Nearest neighbors' data points:")
print(nearest_neighbors)

Output:

[1] "Indices of the nearest neighbors:"
     [,1] [,2] [,3]
[1,]   67   12   66

[1] "Distances to the nearest neighbors:"
          [,1]      [,2]      [,3]
[1,] 0.2921199 0.3538839 0.5632516

[1] "Nearest neighbors' data points:"
          x        y
67 5.896420 6.273139
12 5.719628 6.215929
66 5.607057 5.596455

Step 5. Visualize the Results

Visualizing the dataset and the k-nearest neighbors can help in understanding the algorithm’s output. Use the ggplot2 library for visualization.

library(ggplot2)

# Plot the dataset
ggplot(data, aes(x = x, y = y)) +
  geom_point(color = "blue") +
  geom_point(data = nearest_neighbors, aes(x = x, y = y), color = "red", size = 3) +
  geom_point(data = point, aes(x = x, y = y), color = "green", size = 3) +
  ggtitle("k-Nearest Neighbors") +
  xlab("Feature 1") +
  ylab("Feature 2")

Output:

Finding k-nearest Neighbor for Only One Point Using R

Conclusion

Finding the k-nearest neighbors for a single point in R is straightforward with the help of the FNN package. This method is useful in various applications, including anomaly detection, recommendation systems, and more. By following the steps outlined in this article, you can efficiently identify the k-nearest neighbors for any given point in your dataset.