Transforming data from flat array into hierarchical structure

Question

I'm looking to do a data transformation from a flat list into a hierarchical structure. How can I accomplish this in a readable way but still acceptable in performance and are there any .NET libraries I can take advantage of. I think this is considered a "facet" in certain terminologies (in this case by Industry).

public class Company
{        
    public int CompanyId { get; set; }
    public string CompanyName { get; set; }
    public Industry Industry { get; set; }
}

public class Industry
{
    public int IndustryId { get; set; }
    public string IndustryName { get; set; }
    public int? ParentIndustryId { get; set; }
    public Industry ParentIndustry { get; set; }
    public ICollection<Industry> ChildIndustries { get; set; }
}

Now let's say I have a List<Company> and I'm looking to transform it into a List<IndustryNode>

//Hierarchical data structure
public class IndustryNode
{
    public string IndustryName{ get; set; }
    public double Hits { get; set; }
    public IndustryNode[] ChildIndustryNodes{ get; set; }
}

So that the resulting object should look like this following after it is serialized:

{
    IndustryName: "Industry",
    ChildIndustryNodes: [
        {
            IndustryName: "Energy",
            ChildIndustryNodes: [
                {
                    IndustryName: "Energy Equipment & Services",
                    ChildIndustryNodes: [
                        { IndustryName: "Oil & Gas Drilling", Hits: 8 },
                        { IndustryName: "Oil & Gas Equipment & Services", Hits: 4 }
                    ]
                },
                {
                    IndustryName: "Oil & Gas",
                    ChildIndustryNodes: [
                        { IndustryName: "Integrated Oil & Gas", Hits: 13 },
                        { IndustryName: "Oil & Gas Exploration & Production", Hits: 5 },
                        { IndustryName: "Oil & Gas Refining & Marketing & Transporation", Hits: 22 }
                    ]
                }
            ]
        },
        {
            IndustryName: "Materials",
            ChildIndustryNodes: [
                {
                    IndustryName: "Chemicals",
                    ChildIndustryNodes: [
                        { IndustryName: "Commodity Chemicals", Hits: 24 },
                        { IndustryName: "Diversified Chemicals", Hits: 66 },
                        { IndustryName: "Fertilizers & Agricultural Chemicals", Hits: 22 },
                        { IndustryName: "Industrial Gases", Hits: 11 },
                        { IndustryName: "Specialty Chemicals", Hits: 43 }
                    ]
                }
            ]
        }
    ]
}

Where "Hits" are the number of companies that fall into that group.

To clarify, I need to transform a List<Company> into a List<IndustryNode> NOT serialize a List<IndustryNode>

What do you mean by efficiency? The most readable and maintainable or the best performing? — CookieOfFortune
– CookieOfFortune, Commented Oct 15, 2013 at 15:29
Sorry I didn't make it clear. It needs to be effecient but I'm willing to make some trade off for readability and maintainability — parliament
– parliament, Commented Oct 15, 2013 at 15:30

Ahmed KRAIEM · Accepted Answer · 2013-10-15 16:21:36Z

1

Try this:

    private static IEnumerable<Industry> GetAllIndustries(Industry ind)
    {
        yield return ind;
        foreach (var item in ind.ChildIndustries)
        {
            foreach (var inner in GetAllIndustries(item))
            {
                yield return inner;
            }
        }
    }

    private static IndustryNode[] GetChildIndustries(Industry i)
    {
        return i.ChildIndustries.Select(ii => new IndustryNode()
        {
            IndustryName = ii.IndustryName,
            Hits = counts[ii],
            ChildIndustryNodes = GetChildIndustries(ii)
        }).ToArray();
    }


    private static Dictionary<Industry, int> counts;
    static void Main(string[] args)
    {
        List<Company> companies = new List<Company>();
        //...
        var allIndustries = companies.SelectMany(c => GetAllIndustries(c.Industry)).ToList();
        HashSet<Industry> distinctInd = new HashSet<Industry>(allIndustries);
        counts = distinctInd.ToDictionary(e => e, e => allIndustries.Count(i => i == e));
        var listTop = distinctInd.Where(i => i.ParentIndustry == null)
                        .Select(i =>  new IndustryNode()
                                {
                                    ChildIndustryNodes = GetChildIndustries(i),
                                    Hits = counts[i],
                                    IndustryName = i.IndustryName
                                }
                        );
    }

untested

edited Oct 15, 2013 at 16:21

answered Oct 15, 2013 at 15:30

Ahmed KRAIEM

10.4k4 gold badges32 silver badges33 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

parliament Over a year ago

distrinctInd.Where(i => i.ParentIndustry == null) doesnt match any elements because the companies never reference any top level Industry elements. I've been trying to make it work otherwise but still am having much difficulty.

Ahmed KRAIEM Over a year ago

Try distinctInd.Where(i => i.ChildIndustries == null || i.ChildIndustries.Count == 0)

CodeChops · Accepted Answer · 2013-10-15 15:31:43Z

0

You are looking for a serializer. MSFT has one that is native to VS, but I like Newtonsofts, which is free. MSFT documentation and examples are here, Newtonsoft documentation is here.

Newtonsoft is free, easy and faster.

answered Oct 15, 2013 at 15:31

CodeChops

2,0781 gold badge21 silver badges28 bronze badges

3 Comments

CodeChops Over a year ago

I really don't like someone giving me a minus one with no reason. If you don't have a reason, don't vote it down.

parliament Over a year ago

I didn't downvote but the answer is not helpful. I'll already be using JSON.NET to serialize but I still need to get it into the proper structure.

CodeChops Over a year ago

That wasn't clear in the original post (as evidenced in half the answers). It sounded like you were looking for performance. Sorry I misunderstood your question. I still think it's lousy to minus one anything and not explain your reason.

Community · Accepted Answer · 2017-05-23 12:05:14Z

0

Try to use json serializer for this purpose. I see that you data structure is OK, this is just a matter of serialization.

var industryNodeInstance = LoadIndustryNodeInstance();

var json = new JavaScriptSerializer().Serialize(industryNodeInstance);

If you want to choose between serializers please see this: http://www.servicestack.net/benchmarks/#burningmonk-benchmarks

LoadIndustryNodeInstance method

Build List<Industry>
Convert IndustryTree = List<IndustryNode>
Implement Tree methods, such Traverse. Try to look at Tree data structure in C#

edited May 23, 2017 at 12:05

CommunityBot

11 silver badge

answered Oct 15, 2013 at 15:44

Alexandr

1,4602 gold badges20 silver badges43 bronze badges

1 Comment

parliament Over a year ago

The question concerns what happens in LoadIndustryNodeInstance(). I have a List<Company> not a List<IndustryNode>

CookieOfFortune · Accepted Answer · 2013-10-15 16:09:24Z

0

Here is some psuedo code that might get you along the way. I create a map/dictionary index and populate it with the company list. Then we extract the top level nodes from the index. Note that there may be edge cases (For example, this index may need to be partially filled initially as it doesn't seem any of your companies ever reference the very top level nodes, so those will have to be filled in some other way).

Dictionary<String, IndustryNode> index = new Dictionary<String, IndustryNode>();

public void insert(Company company)
{ 
    if(index.ContainsKey(company.Industry.IndustryName))
    {
        index[company.Industry.IndustryName].hits++;
    }
    else
    {
        IndustryNode node = new IndustryNode(IndustryName=company.Industry, Hits=1);
        index[node.IndustryName] = node;
        if(index.ContainsKey(company.Industry.ParentIndustry.IndustryName))
        {
            index[company.Industry.ParentIndustry.IndustryName].ChildrenIndustries.Add(node);
        }
    }    
}

List<IndustryNode> topLevelNodes = index
    .Where(kvp => kvp.Item.ParentIndustry == null)
    .ToList(kvp => kvp.Item);

edited Oct 15, 2013 at 16:09

answered Oct 15, 2013 at 16:04

CookieOfFortune

14k8 gold badges45 silver badges58 bronze badges

4 Comments

Ahmed KRAIEM Over a year ago

This solution will not take into account children of children of an industry if it is not affected to a company.

CookieOfFortune Over a year ago

@AhmedKRAIEM True, those would have to be inserted initially.

parliament Over a year ago

Thanks for the answer, if this method took an Industry instead where could recursion be applied to handle the children of children case?

CookieOfFortune Over a year ago

Can you explain further? The way the data is currently presented, recursion isn't immediately usable. Eg. You could search a tree via recursion but that won't be different from a linear search since there's no stated search order.

Collectives™ on Stack Overflow

Transforming data from flat array into hierarchical structure

4 Answers 4

2 Comments

3 Comments

1 Comment

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

4 Answers 4

2 Comments

3 Comments

1 Comment

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related