We Keep Coding

Linq Group By multiple fields and Flatten the list

I've got the following data
title | useful
ttitle1 | Yes
ttitle1 | Yes
ttitle1 | No
ttitle2 | Yes
I would like to group the above data and flatten it so I get the following result:
Title | Useful Count | Not Useful Count
tttitle1 | 2 | 1
tttitle2 | 1 | 0
Tried this, but it does not produce the correct result:
var query = (from r in ratings
group r by new { r.ArticleTitle, r.Useful } into results
group results by new { results.Key.ArticleTitle } into results2
from result in results2
select new
{
Title = result.Key.ArticleTitle,
Yes = result.Select(i => i.Useful).Count(),
No = result.Select(i => i.Useful == false).Count()
});
Any help?

It seems to me that the only problem is that you're grouping twice. I'd expect this to work:
var query = from rating in ratings
group rating by rating.ArticleTitle into g
select new
{
Title = g.Key,
Yes = g.Count(r => r.Useful),
No = g.Count(r => !r.Useful)
};
Or not in query expression form:
var query = ratings.GroupBy(r => r.ArticleTitle,
(key, rs) => new
{
Title = key,
Yes = rs.Count(r => r.Useful),
No = rs.Count(r => !r.Useful)
});

You don't need to group twice to get the desired result. One Grouping would be fine:
var query = (from r in ratings
group r by new { r.ArticleTitle } into g
from result in groups
select new
{
Title = result.Key,
Yes = result.Select(i => i.Useful).Count(),
No = result.Select(i => !i.Useful).Count()
});

LINQ sum not performed when values are identical

I need to sum elements of same type starting from 2 LINQ queries.
Below is my code:
var query1 = from d in _contextProvider.Context.Documents
where d.TransportId == transportId
group d by d.Type
into dg
select new { DocumentType = dg.Key.ToString(), DocumentCount = dg.Count() };
var query2 = from n in _contextProvider.Context.NotificationDocuments
where n.TransportId == transportId
group n by n.TransportId
into nd
select new { DocumentType = "Notification", DocumentCount = nd.Count() };
var query_collapsed = query1.Union(query2)
.GroupBy(p => new { DocumentType = p.DocumentType })
.Select(g => new DocumentCounters() { DocumentType = g.Key.DocumentType, DocumentCount = g.Sum(p => p.DocumentCount) });
Example: below let's analyse values for DocumentType equals to Notification.
Values of query1:
Values of query2:
The collapsed query :
That's correct: 1 + 2 = 3
The problem: I noticed that whenever the count for Notification in query1 is equals to the count for Notification in query2, then the sum is not performed.
Example:
2 + 2 = 2
or
3 + 3 = 3
Any ideas ?

LINQ Union will remove duplicate entries. If you want to merge the two sequences you can use Concat like so:
var query_collapsed = query1.Concat(query2)
.GroupBy(p => new { DocumentType = p.DocumentType })
.Select(g => new DocumentCounters() { DocumentType = g.Key.DocumentType, DocumentCount = g.Sum(p => p.DocumentCount) });

Select Last non null-able item per product?

Let's say I have,
class Product
{
public int Id {get; set;}
public string Name {get; set;}
public int Order {get; set;}
}
and my data have,
products[0] = new Product { Id = 1, Name = "P1", Order = 1 };
products[1] = new Product { Id = 1, Name = "P2", Order = 2 };
products[2] = new Product { Id = 1, Name = null, Order = 3 };
products[3] = new Product { Id = 2, Name = "P3", Order = 4 };
products[4] = new Product { Id = 2, Name = null, Order = 5 };
products[5] = new Product { Id = 2, Name = null, Order = 6 };
What I need is the last(order by Order desc) non-nullable value of Name per Product.Id. So my final output will look like,
items[0] = new { Id = 1, Name = "P2"};
items[1] = new { Id = 2, Name = "P3"};
If Id=1, I have 3 Names (P1, P2, null) and non-nullable Names (P1, P2) but last one is P3.

This should get the last products in order.
var lastOrders = products
.Where(x => x.Name != null) // Remove inapplicable data
.OrderBy(x => x.Order) // Order by the Order
.GroupBy(x => x.Id) // Group the sorted Products
.Select(x => x.Last()); // Get the last products in the groups

var result = products
.GroupBy(p => p.Id)
.Select(g => g.OrderBy(x => x.Order).Last(x => x.Name != null));

this will give you your desired output:
products.GroupBy(p => p.Id)
.Select(g => g.OrderByDescending(gg => gg.Name)
.Where(gg => gg.Name != null)
.Select(gg => new { gg.Id, gg.Name })
.First());

The task can be solved using the following Linq statement.
var Result = products.OrderBy().Where( null != iProduct.Name ).First();
This requires products to contain at least one item where Name is null, otherwise an Exception will be thrown. Alternatively,
var Result = products.OrderBy().Where( null != iProduct.Name ).FirstOrDefault();
will return null if products contains no such item.

Try with :
var expectedProduct =products.Where(p => p.Id != null).OrderByDescending(p => p.Order).GroupBy(p => p.Id).Last()

C# Linq Average

I have a table with data similar to below:
Group TimePoint Value
1 0 1
1 0 2
1 0 3
1 1 3
1 1 5
I want to project a table as such:
Group TimePoint AverageValue
1 0 2
1 1 4
EDIT: The data is in a datatable.
Anybody any ideas how this can be done with LINQ or otherwise?
Thanks.

You need to perform Group By
The linq you need is something like:
var query = from item in inputTable
group item by new { Group = item.Group, TimePoint = item.TimePoint } into grouped
select new
{
Group = grouped.Key.Group,
TimePoint = grouped.Key.TimePoint,
AverageValue = grouped.Average(x => x.Value)
} ;
For more Linq samples, I highly recommend the 101 Linq samples page - http://msdn.microsoft.com/en-us/vcsharp/aa336747#avgGrouped

Here's a more function-oriented approach (the way I prefer it). The first line won't compile, so fill it in with your data instead.
var items = new[] { new { Group = 1, TimePoint = 0, Value = 1} ... };
var answer = items.GroupBy(x => new { TimePoint = x.TimePoint, Group = x.Group })
.Select(x => new {
Group = x.Key.Group,
TimePoint = x.Key.TimePoint,
AverageValue = x.Average(y => y.Value),
}
);

You can do:
IEnumerable<MyClass> table = ...
var query = from item in table
group item by new { item.Group, item.TimePoint } into g
select new
{
g.Key.Group,
g.Key.TimePoint,
AverageValue = g.Average(i => i.Value)
};

Assuming a class like this:
public class Record
{
public int Group {get;set;}
public int TimePoint {get;set;}
public int Value {get;set;}
}
var groupAverage = from r in records
group r by new { r.Group, r.TimePoint } into groups
select new
{
Group = groups.Key.Group,
TimePoint = groups.Key.TimePoint,
AverageValue = groups.Average(rec => rec.Value)
};

Applying a filter to subsequences of a sequence using Linq

If I have a List<MyType> as so, with each line representing an item in the collection:
{{ Id = 1, Year = 2010 },
{ Id = 1, Year = 2009 },
{ Id = 1, Year = 2008 },
{ Id = 2, Year = 2010 },
{ Id = 2, Year = 2009 },
{ Id = 2, Year = 2008 }}
I wish to retrieve a collection from this collection of the most recent item for each Id. What will the Linq for this look like?
Desired output:
{{ Id = 1, Year = 2010 },
{ Id = 2, Year = 2010 }}
I have a naiive implementation using a second list variable and a foreach loop, but it's inefficient.
//naiive implementation "p-code"
//...
var mostRecentItems = new List<MyType>();
var ids = collection.Select(i => i.Id).Distinct();
foreach(var id in ids)
{
mostRecentItems.Add(collection.Where(i => i.Id == id).OrderByDescending().First);
}
return mostRecentItems;

Most simply:
var mostRecentById = from item in list
group item by item.Id into g
select g.OrderByDescending(x => x.Year).First();

Group by id, then select the first item in each group ordered in a descending fashion.
var mostRecentItems = collection.GroupBy( c => c.Id )
.Select( g => g.OrderByDescending( i => i.Year ).First() );

or more simply still:
var result = list
.GroupBy(i => i.Id)
.Select(g => new {Id = g.Key, Year = g.Max(y => y.Year)});