Half-assed apriori attempt.

gistfile1.d

module apriori;

import std.stdio;
import std.string;
import std.conv;

bool contains(A, E)(A array, E element) {
    foreach(E e; array) {
        if(e == element) return true;
    }
    return false;
}

void main(string[] args) {
    if(args.length < 2) {
        writeln("USAGE: ", args[0], " supportThreshold");
        return;
    }

    auto supportProc = to!double(args[1]);

    string[][] input;
    ulong[string] itemSets;
    ulong allRules;
    string[][string] rules;

    // Build 1-item sets
    foreach(line; stdin.byLine()) {
        auto words = line.idup.strip().split(",");
        input ~= words;

        foreach(word; words) {
            itemSets[word]++;
        }
    }

    double supportThreshold = supportProc * input.length;

    // Build frequent 1-item sets

    ulong[string] tmp;
    foreach(key, support; itemSets) {
        if(support > supportThreshold) tmp[key] = support;
    }
    itemSets = tmp;

    // Build frequent 2-item sets

    foreach(key1; itemSets.keys) {
        foreach(key2; itemSets.keys) {
            if(key1 != key2) {
                ulong support;

                foreach(line; input) {
                    if(contains(line, key1) && contains(line, key2)) support++;
                }

                if(support > supportThreshold) {
                    allRules += 2;
                    rules[key1] ~= key2;
                    rules[key2] ~= key1;
                }
            }
        }
    }

    // Etc...

    // Build index mapping.
    ulong[string] indexMapping;
    foreach(i, key; itemSets.keys) {
        indexMapping[key] = i;
    }

    // JSON output as a graph.
    writeln("{\"nodes\": [");

    foreach(name, support; itemSets) {
        writeln("\t{");
        writeln("\t\t\"id\": ", indexMapping[name], ",");
        writeln("\t\t\"label\": \"", name, "\",");
        writeln("\t\t\"support\": ", (cast(double) support) / input.length);
        write("\t}");
        if(indexMapping[name] < indexMapping.length-1) write(",");
        writeln();
    }

    writeln("], \"edges\": [");

    ulong i;
    foreach(name; itemSets.keys) {
        if(name !in rules) continue;

        foreach(target; rules[name]) {
            writeln("\t{");
            writeln("\t\t\"id\": ", i++, ",");
            writeln("\t\t\"source\": ", indexMapping[name], ",");
            writeln("\t\t\"target\": ", indexMapping[target]);
            write("\t}");
            if(i < allRules) write(",");
            writeln();
        }
    }

    writeln("]}");
}