Apriori Class

public class Apriori : Apriori<int>

Public Class Apriori
	Inherits Apriori(Of Integer)

// Example from https://en.wikipedia.org/wiki/Apriori_algorithm

// Assume that a large supermarket tracks sales data by stock-keeping unit
// (SKU) for each item: each item, such as "butter" or "bread", is identified 
// by a numerical SKU. The supermarket has a database of transactions where each
// transaction is a set of SKUs that were bought together.

// Let the database of transactions consist of following itemsets:

SortedSet<int>[] dataset =
{
    // Each row represents a set of items that have been bought 
    // together. Each number is a SKU identifier for a product.
    new SortedSet<int> { 1, 2, 3, 4 }, // bought 4 items
    new SortedSet<int> { 1, 2, 4 },    // bought 3 items
    new SortedSet<int> { 1, 2 },       // bought 2 items
    new SortedSet<int> { 2, 3, 4 },    // ...
    new SortedSet<int> { 2, 3 },
    new SortedSet<int> { 3, 4 },
    new SortedSet<int> { 2, 4 },
};

// We will use Apriori to determine the frequent item sets of this database.
// To do this, we will say that an item set is frequent if it appears in at 
// least 3 transactions of the database: the value 3 is the support threshold.

// Create a new a-priori learning algorithm with support 3
Apriori apriori = new Apriori(threshold: 3, confidence: 0);

// Use the algorithm to learn a set matcher
AssociationRuleMatcher<int> classifier = apriori.Learn(dataset);

// Use the classifier to find orders that are similar to 
// orders where clients have bought items 1 and 2 together:
int[][] matches = classifier.Decide(new[] { 1, 2 });

// The result should be:
// 
//   new int[][]
//   {
//       new int[] { 4 },
//       new int[] { 3 }
//   };

// Meaning the most likely product to go alongside the products
// being bought is item 4, and the second most likely is item 3.

// We can also obtain the association rules from frequent itemsets:
AssociationRule<int>[] rules = classifier.Rules;

// The result will be:
// {
//     [1] -> [2]; support: 3, confidence: 1, 
//     [2] -> [1]; support: 3, confidence: 0.5, 
//     [2] -> [3]; support: 3, confidence: 0.5, 
//     [3] -> [2]; support: 3, confidence: 0.75, 
//     [2] -> [4]; support: 4, confidence: 0.66, 
//     [4] -> [2]; support: 4, confidence: 0.8, 
//     [3] -> [4]; support: 3, confidence: 0.75, 
//     [4] -> [3]; support: 3, confidence: 0.6 
// };

// Example from http://www3.cs.stonybrook.edu/~cse634/lecture_notes/07apriori.pdf

// For this example, we will consider a database consisting of 9 transactions:
// - Minimum support count required will be 2 (i.e. min_sup = 2 / 9 = 22%);
// - Minimum confidence required should be 70%;

// The dataset of transactions can be as follows:
string[][] dataset =
{
    new string[] { "1", "2", "5" },
    new string[] { "2", "4" },
    new string[] { "2", "3" },
    new string[] { "1", "2", "4" },
    new string[] { "1", "3" },
    new string[] { "2", "3" },
    new string[] { "1", "3" },
    new string[] { "1", "2", "3", "5" },
    new string[] { "1", "2", "3" },
};

// Create a new A-priori learning algorithm with the requirements
var apriori = new Apriori<string>(threshold: 2, confidence: 0.7);

// Use apriori to generate a n-itemset generation frequent pattern
AssociationRuleMatcher<string> classifier = apriori.Learn(dataset);

// Generate association rules from the itemsets:
AssociationRule<string>[] rules = classifier.Rules;

// The result should be:
// {
//   [5]   -> [1];   support: 2, confidence: 1,
//   [5]   -> [2];   support: 2, confidence: 1,
//   [4]   -> [2];   support: 2, confidence: 1,
//   [5]   -> [1 2]; support: 2, confidence: 1,
//   [1 5] -> [2];   support: 2, confidence: 1,
//   [2 5] -> [1];   support: 2, confidence: 1,
// }