Selecting sum of first and last rows in related table - sql

Question

In my table I have these three tables:
Teams:
Id GUID UNIQUE PRIMARY
Name NVARCHAR
Participants:
Id GUID UNIQUE PRIMARY
FirstName NVARCHAR
[....]
TeamId GUID
ParticipantData:
Id GUID UNIQUE PRIMARY
Weight FLOAT
Date DATETIME
ParticipantId GUID
TeamId GUID
What I need is an SQL query that gives me all columns from Teams AND:
The sum of the first (order by Date) entries in ParticipantData of participants in the team (TeamId)
The sum of the last (ordered by Date) entries in ParticipantData of participants in the team (TeamId)
Explanation:
I have lots of participants (team members) reporting their weight with some interval (Weight + Date). What I'm trying to accomplish is to calculate the weight loss of all the team members.
On 2019-01-03 Participant 1 reports Weight 78
On 2019-01-06 Participant 1 reports Weight 75
On 2019-01-04 Participant 2 reports Weight 86
On 2019-01-07 Participant 2 reports Weight 83
I need the query to get SumOfFirstWeights (78 + 86) = 164
I need the query to get SumOfLastWeights (75 + 83) = 158
Which gives me a weight loss of 6.
I've tried many combinations of:
Select *,
(SELECT TOP (1) Sum(Weight)
FROM ParticipantData WHERE (TeamId = Teams.Id)
GROUP BY ParticipantId
)
ORDER BY Date As SumOfFirstWeights
From Teams

Answer 1

Your problem is some kind of greatest/lowest per group, moreover, you want a sum of these values.
select t.id, t.name, sum(t1.weight), sum(t2.weight)
from teams t
left join
(
select pd.teamid, pd.participantid, pd.weight
from ParticipantData pd
join
(
select teamid, participantid, min(date) min_date
from ParticipantData
group by teamid, participantid
) t on pd.teamid = t.teamid and
pd.participantid = t.participantid and
pd.date = t.min_date
) t1 on t.id = t1.teamid
left join
(
select pd.teamid, pd.participantid, pd.weight
from ParticipantData pd
join
(
select teamid, participantid, max(date) max_date
from ParticipantData
group by teamid, participantid
) t on pd.teamid = t.teamid and
pd.participantid = t.participantid and
pd.date = t.max_date
) t2 on t1.teamid = t2.teamid and
t1.participantid = t2.participantid
group by t.id, t.name

Answer 2

You want to find the first and last row per team. You can use analytic functions for this. I'm using ROW_NUMBER which gives me exactly one record, even if a participant has two entries on the same day.
with pd as
(
select
participantid,
weight,
row_number() over (partition by participantid order by date) as one_is_first,
row_number() over (partition by participantid order by date desc) as one_is_last
from participantdata
)
, team_weights as
(
select
p.teamid,
sum(case when pd.one_is_first = 1 then weight end) as first_weight,
sum(case when pd.one_is_last = 1 then weight end) as last_weight,
sum(case when pd.one_is_last = 1 then weight end) -
sum(case when pd.one_is_first = 1 then weight end) as difference
from pd
join participant p on p.id = pd.participantid
where (pd.one_is_first = 1 or pd.one_is_last = 1)
group by p.teamid
)
select *
from teams
join team_weights on where team_weights.team_id = teams.id
order by teams.id;