How to find similar objects in a collection

Question

I want to find similar objects in a collection depending on a Method I implement

for example this sample class:

class myObj
{
    public int Data1 { get; set; }

    public int Data2 { get; set; }

    public int Data3 { get; set; }        
}

then implement Similar method in the class:

public bool Similar(myObj obj)
{
    if (obj.Data1 == this.Data1 && obj.Data2 == this.Data2)
        return true;
    return false;
}

now I have this collection:

List<myObj> items = new List<myObj>();
// none similar
items.Add(new myObj() { Data1 = 1, Data2 = 2, Data3 = 4 });
items.Add(new myObj() { Data1 = 2, Data2 = 3, Data3 = 18 });
items.Add(new myObj() { Data1 = 3, Data2 = 4, Data3 = 75 });
items.Add(new myObj() { Data1 = 4, Data2 = 2, Data3 = 3 });
//similar
items.Add(new myObj() { Data1 = 5, Data2 = 26, Data3 = 97 });
items.Add(new myObj() { Data1 = 5, Data2 = 26, Data3 = 37 });
items.Add(new myObj() { Data1 = 10, Data2 = 45, Data3 = 47 });
items.Add(new myObj() { Data1 = 10, Data2 = 45, Data3 = 19 });

to get the similar objects I did this:

private static List<myObj> GetSimilars(List<myObj> items)
{
    List<myObj> similars = new List<myObj>();
    while (items.Count > 0)
    {
        var q = (from c in items
                 where c.Similar(items[0])
                 select c).ToList();

        if (q.Count > 1)
        {
            similars.AddRange(q);
            foreach (var obj in q)
                items.Remove(obj);
        }
        else
            items.Remove(items[0]);
    }    
    return similars;
}

is there a better way to do that?

Answer 1

How about making this class, which is reusable.

public class MyObjSimilarity : EqualityComparer<myObj>
{
    public override bool Equals(myObj a, myObj b)
    {
        if (obj.Data1 == this.Data1 && obj.Data2 == this.Data2)
        {
            return true;
        }

        return false;
    }

    public override int GetHashCode(myObj o)
    {
        int hash = 17;
        hash = hash * 23 + o.Data1.GetHashCode();
        hash = hash * 23 + o.Data2.GetHashCode();
        return hash;
    }
}

You could use like this,

var similarity = new MyObjSimilarity();
items.Where(o => similarity.Equals(o, w));

or pass into the constructor of a dictionary,

var similarity = new MyObjSimilarity();
var lookup = new Dictionary<myObj, string>(similarity);

or in a GroupBy

var similarity = new MyObjSimilarity();
items.GroupBy(o => o, o => o, similarity);

or like the other answer

var similarity = new MyObjSimilarity();
items.GroupBy(
    o => o,
    o => new { Instance = o, Count = Count(o) },
    similarity);

or in other framework friendly places.

Answer 2

try this:

private static List<myObj> GetSimilars(List<myObj> items)
{
    return items.SelectMany(x => items.Where(z => x != z && x.Similar(z))).ToList();
}

or if you prefer this:

private static List<myObj> GetSimilars(List<myObj> items)
    {
        var result = from x in items
                     from y in items
                     where x != y && x.Similar(y)
                     select x;

        return result.ToList();
    }

Answer 3

You could do that all with Linq's GroupBy and SelectMany:

var similarGroups = from i in items
                    group i by new { i.Data1, i.Data2 } into D1D2Group
                    where D1D2Group.Count() > 1
                    select D1D2Group;

foreach (var grp in similarGroups)
    Console.WriteLine("DataGroup:{0}/{1} Count:{2}"
               , grp.Key.Data1
               , grp.Key.Data2
               , grp.Count());

If you want to flatten the groups to a List<myObj> like your GetSimilars:

 List <myObj> similars = similarGroups.SelectMany(g => g).ToList();

Answer 4

return
    items.Where(w => items.Count(c => c.Similar(w)) > 1)
    /* add .Distinct() optional*/
    .ToList();