How to write this LINQ Query

Question

Hi i have a datatable with following fields

DAT_START
GROUPBY
TXT_LATITTUDE
TXT_LONGITUDE
INT_DIRECTION
INT_CALL_DATA_TYPE
LNG_DURATION

And following is the LINQ query i am using

var data = (from r in dt.AsEnumerable()
where ((r.Field<DateTime>("DAT_START").TimeOfDay.Hours < 20) && (r.Field<DateTime>("DAT_START").TimeOfDay.Hours >= 4))
group r by new { CID = r["GroupBy"], CLatitude = r["TXT_LATITUDE"], CLongitude = r["TXT_LONGITUDE"],CDirection = r["INT_DIRECTION"], 
CCallType = r["INT_CALL_DATA_TYPE"],CDuration = r["LNG_DURATION"] }
into groupedTable
select new
{
   CellID = groupedTable.Key.CID,
   CallCount = groupedTable.Count(),
   Longitude = groupedTable.Key.CLongitude,
   Latitude = groupedTable.Key.CLatitude,
   Direction = groupedTable.Key.CDirection,
   CallType = groupedTable.Key.CCallType,
   Duration = groupedTable.Key.CDuration
}).OrderByDescending(s => s.CallCount);

It gives me result like this

CellID = 4057,CallCount = 84,Longitude = "",Latitude = "",Direction = "Incoming",CallType = "Voice",Duration = 50
CellID = 4057,CallCount = 8,Longitude = "",Latitude = "",Direction = "Outgoing",CallType = "Voice",Duration =97
CellID = 4057,CallCount = 56,Longitude = "",Latitude = "",Direction = "Incoming",CallType ="SMS" ,Duration = 0
CellID = 4057,CallCount = 41,Longitude = "",Latitude = "",Direction = "Outgoing",CallType = "SMS",Duration = 0

Now i want result like this

CellID = 4057, TotalCommCount = 204, TotalDuration = 147, INSMSCount = 56,OutSMSCount = 41, INVoiceCount = 84,OutVoiceCount = 8,InVoiceDuration =50,OutVoiceDuration = 47

How can i do this. I am struck over here..

Answer 1

It looks like you're grouping by far too much at the moment, which is why you're getting multiple rows. I suspect you want something like this:

from r in dt.AsEnumerable()
where r.Field<DateTime>("DAT_START").TimeOfDay.Hours < 20 && 
      r.Field<DateTime>("DAT_START").TimeOfDay.Hours >= 4
group r r["GroupBy"] into g         
select new
{
    CellID = g.Key,
    TotalCommCount = g.Count(),
    TotalDuration = g.Sum(r => r.Field<long>("LNG_DURATION")),
    InSMSCount = g.Count(r => r.Field<string>("DIRECTION") == "Incoming" &&
                              r.Field<string>("CALL_TYPE") == "SMS"),
    OutSMSCount = g.Count(r => r.Field<string>("DIRECTION") == "Outgoing" &&
                               r.Field<string>("CALL_TYPE") == "SMS"), 
    InVoiceCount = g.Count(r => r.Field<string>("DIRECTION") == "Incoming" &&
                                r.Field<string>("CALL_TYPE") == "Voice"),
    OutVoiceCount = g.Count(r => r.Field<string>("DIRECTION") == "Outgoing" &&
                                 r.Field<string>("CALL_TYPE") == "Voice"),
    InVoiceDuration = g.Where(r => r.Field<string>("DIRECTION") == "Incoming" &&
                                r.Field<string>("CALL_TYPE") == "Voice")
                       .Sum(r => r.Field<long>("DURATION"))
    OutVoiceDuration = g.Where(r => r.Field<string>("DIRECTION") == "Outgoing" &&
                                    r.Field<string>("CALL_TYPE") == "Voice"),
                       .Sum(r => r.Field<long>("DURATION"))
 } into summary
 order by summary.TotalCommCount descending
 select summary;

Answer 2

I suggest you create a view in your SQL database where you can simply aggregate what you need and then you use that view in your Linq query - this will perform fast and it is easy to write.

Sketch of design:

Since the description of your requirements is not totally clear, I have made assumptions. Create a SQL view such as (it's just to show the idea, change it according to your needs):

CREATE VIEW [dbo].[vSumDurations]
AS 
select  c.CellId, j1.SumDuration1 as TotalDuration, 
        j2.SumDuration2 as GroupedDuration, j1.CallCount  
from (select distinct t.CID as CellId from [dbo].[YourTable] t) c
left join (select t1.CID as CellId, sum(LNG_Duration) as SumDuration1, 
            count(*) as CallCount from [dbo].[YourTable] t1 
            Group By t1.CID) j1 
on c.CellId=j1.CellId           
left join (select t2.CID as CellId, sum(LNG_Duration) as SumDuration2 
            from [dbo].[YourTable] t2 
            Group by t2.CID, t2.Direction, t2.CallType) j2 
on c.CellId=j2.CellId

(You know you can easily add this view to your *.EDMX by using "update model from database" in the context menu of the entity diagram page and then tick-mark the view vSumDurations in the "tables/views" section of the dialog which pops up.)

After this preparation your Linq query is very simple because everything is already done in the view. Hence, the code looks like:

var dc = this; // in LinqPad you can set your data context
var data = (from d in dc.vSumDurations select d).OrderByDescending(s => s.CallCount);
data.Dump();

Note: The example is made for LinqPad - you will need a different data context in your real code. Linqpad does not need dc , but it is easier if you declare it for convenience reasons because in your real code you have to provide it.