I am trying to find an efficient way to compare a given array that represents the Quality of Service of a Client with the QOS of a Two Web Services and pick up the most similar to the given QOS.
int[] arrayCompare(int[] clientQOS){ int[] service1QOS=new int[]{3,1,4}; int[] service2QOS=new int[]{1,3,3}; //I want to compare clientQOS with S1 QOS and S2 QOS return mostSimilarArray } Loop through the source array for each target, incrementing a counter by the delta (positive difference between the values). The smallest delta sum is the closest match.
Caveats
This should be true if you consider numerically closer to be more similar, and the scale between different values are equal.
I'd make a custom object instead of array and implement interface Comparable. This way you can encapsulate algorithm of comparison and reuse it later in other places.
There are a lot of method for calculate array "similarity".
Here the Golang implementation:
// SimilarityPreCheck is delegated to verify that the given array have the correct size func SimilarityPreCheck(a, b []float64) bool { if len(a) == 0 || len(b) == 0 { log.Println("CosineSimilarity | Nil input data") return false } if len(a) != len(b) { log.Printf("CosineSimilarity | Input vectors have different size") return false } return true } // CosineSimilarity is delegated to calculate the Cosine Similarity for the given array func CosineSimilarity(a, b []float64) float64 { if !SimilarityPreCheck(a, b) { return -1 } // Calculate numerator var numerator float64 for i := range a { numerator += a[i] * b[i] } // Caluclate first term of denominator var den1 float64 for i := range a { den1 += math.Pow(a[i], 2) } den1 = math.Sqrt(den1) // Caluclate second term of denominator var den2 float64 for i := range b { den2 += math.Pow(b[i], 2) } den2 = math.Sqrt(den2) result := numerator / (den1 * den2) return result } // EuclideanDistance is delegated to calculate the euclidean distance for the given array func EuclideanDistance(v1, v2 []float64) float64 { if !SimilarityPreCheck(v1, v2) { return -1 } var euclidean float64 for i := range v1 { euclidean += math.Pow(v1[i]-v2[i], 2) } return math.Sqrt(euclidean) } // ManhattanDistance is delegated to calculate the Manhattan norm for the given array func ManhattanDistance(v1, v2 []float64) float64 { if !SimilarityPreCheck(v1, v2) { return -1 } var taxicab float64 for i := range v1 { taxicab += math.Abs(v2[i] - v1[i]) } return taxicab } 