|
SELECT ID, MIN(EmailAddress)
FROM TEST
GROUP BY ID
Bastard Programmer from Hell 
|
|
|
|
|
Are you sure you've got that query right?
|
|
|
|
|
Is there are reason that you want me to doubt the post?
Bastard Programmer from Hell
|
|
|
|
|
Look at the field in the min as a hint.
|
|
|
|
|
I did. He wants the ID's unique, and the first mailadress for that ID.
Did I miss something? 
Bastard Programmer from Hell
|
|
|
|
|
Hi, everybody
I need to conduct a large amount of data analysis on database. Could anyone recommend an interactive application for data analysis?
The requirements are:
1. Able to cope with the unexpected requirement rapidly.
2. Able to perform further computations on results interactively (base on the mass data).
3. Easy to confront even a large amount of complex computations
What would you great expert recommend?
Thanks in advance.
|
|
|
|
|
Microsoft Excel and Sql Server. Those are the most-used all-purpose tools to do what you want. You can adapt using VBScript if Excel if those requirements change.
Most data-mining applications are written to mine for specific stuff.
Bastard Programmer from Hell
|
|
|
|
|
Thank you for reply.
SQl lacks stepwise computation ablility, so It is not a ideal solution. What I mean about stepwise is:
1. In excel, I can write number "10" in A1 cell, then formula "=A1*5" in A2 cell. then formula "=A2+2" in A3 cell. A1->A2->A3, It's stepwise.
2. When I change 10 to 9 in A1, then result in A2 and A3 will changed automatic.
3. an example: a. to select out the 10 categories of best sellers b. as a further computation on the basis of result from a., to select out the top 20 products from each category, c. as a further comparison with that of the last year based on the result from a.
let's call SQL_A,SQL_B,SQL_C as SQL statements for a,b,c. when I changed SQL_A (for example change 10 categories to 9 categories ), I must change SQL_B and SQL_C. why? because SQL_B and SQL_C is based on SQL_A. that is, SQL_B may like:
with SQL_A as A
selet.......
Excel has another problem, It is too simple to process mass data analysis. for example, to compute the product whose annual sales values are all among the top 100.
data structure( sales table's fields): productID, time, value .
the sql solution is:
WITH sales1 AS (
SELECT productID, YEAR(time) AS year, SUM(value) AS value1
FROM sales
GROUP BY productID, YEAR(time)
)
SELECT productID
FROM (
SELECT productID
FROM (
SELECT productID,RANK() OVER(PARTITION BY year ORDER BY value1 DESC) rankorder
FROM sales1 ) T1
WHERE rankorder<=100) T2
GROUP BY productID
HAVING COUNT(*)=(SELECT COUNT(DISTINCT year ) FROM sales1)
above code is hard to write in Excel.
So what I want is some like SQL + Excel, do you have some idea?
|
|
|
|
|
bestbird7788 wrote: So what I want is some like SQL + Excel, do you have some idea?
Use Sql and Excel.
bestbird7788 wrote: above code is hard to write in Excel.
It is. I'd suggest writing it in Sql, as a stored procedure.
bestbird7788 wrote: Excel has another problem, It is too simple to process mass data analysis.
It's one of the most advanced spreadsheets that I know. Alternatively, you could check out DMTL[^].
Bastard Programmer from Hell
|
|
|
|
|
you are so funny 
I mean SQL+Excel in one tools, not one of them or both of them.
I will check DMTL, It seems like what I want. but I'm not sure does DMTL have stepwise ablility like excel? It seems just a code library. I should take a deeper research.
BTW, does DMTL support computing data with multiple dataset, just as join statement in SQL. Under my impression, other data mining tools can only work on one dataset.
thank you in advance!
|
|
|
|
|
bestbird7788 wrote: you are so funny
I was not joking.
bestbird7788 wrote: I mean SQL+Excel in one tools, not one of them or both of them.
You can use Sql Server to prepare the dataset (and yes, it can operate on the previous record), and Excel to visualize that data.
bestbird7788 wrote: I'm not sure does DMTL
Neither do I. Never needed anything beyond the two tools I already named.
Bastard Programmer from Hell
|
|
|
|
|
hi,thank you for your suggestion.
this line:
You can use Sql Server to prepare the dataset (and yes, it can operate on the previous record), and Excel to visualize that data.
------------------------------------
look at this scene:
1. I use sql to get some data, filled them in Excel, called it dataset1.
2. In Excel, I made some computation, then I got dataset2.
3. I need join the dataset2 with a table in database to generate dataset3.
how to do? creat a new table in database, fill dataset2 to database? and run join statment? then fill dataset3 to Excel? No, It's a big trouble.
so I need a tools like SQL+Excel.
|
|
|
|
|
bestbird7788 wrote: how to do?
Create a linked server to both Excell files, and do a SELECT with a UNION into your Sql Server.
bestbird7788 wrote: No, It's a big trouble.
so I need a tools like SQL+Excel.
Good luck
Bastard Programmer from Hell
|
|
|
|
|
bestbird7788 wrote: What would you great expert recommend?
That you adjust your mind set.
You seem to think that you have something that is easy yet solves anything. Which isn't possible.
The reason programming languages exist is because complex(=many steps) problems can only be solved using complex implementations.
So what you can do is one of the following
1. Find an existing application (like excel) and limit your problems to ones that work in that.
2. Use a programming language and solve anything.
3. Use a mix of 1 and 2 based on the needs of the problem.
Obviously there are disadvantages to all but that does cover the possibilities.
|
|
|
|
|
Thank you for your advice.
I know you are right, there is no easy way to solve my problem.
I think SQL is more easy than programming languages like VB and C++ on common mass data computation, so I imagined if there is a tool that is more easy than SQL on multiple steps computation of mass data. So I posted this post and wanted someone who knows more than me.
DMTL maybe a good choice, I will make a research on it.
|
|
|
|
|
If money is no issue, you can always get Qlikview[^].
|
|
|
|
|
thank you
I will check it.
And I have 3 quesiton about Qlikview ,and I will appreciate if you give me some tips.
1.can Qlikview solve the problem above:to compute the product whose annual sales values are all among the top 100.
MSSQL data structure( sales table's fields): productID, time, value
SQL solution is as below:
---------------------------------------------------------
WITH sales1 AS (
SELECT productID, YEAR(time) AS year, SUM(value) AS value1
FROM sales
GROUP BY productID, YEAR(time)
)
SELECT productID
FROM (
SELECT productID
FROM (
SELECT productID,RANK() OVER(PARTITION BY year ORDER BY value1 DESC) rankorder
FROM sales1 ) T1
WHERE rankorder<=100) T2
GROUP BY productID
HAVING COUNT(*)=(SELECT COUNT(DISTINCT year ) FROM sales1)
2.does Qlikview have stepwise ability. What I mean is like: step1. filter data. step2. group data on step1 . step3. filter data again on step2. the most important is: step4. sorting data on the result of step1 as a sub step. It's like what Excel does.
3.dees Qlikview support multiple datasets computation without SQL. what I mean is like: join statements in SQL .
thank you for your reply.
|
|
|
|
|
Hi this is Chandra. i have one issue with Sql Stored procedure.
Please find the query below.
declare @Measure varchar(50)
declare @Time varchar(50)
declare @Manufacturer varchar(50)
declare @CategoryID int
SET @Measure ='Value'
SET @Time ='4 Weeks&'
SET @Manufacturer ='Jti Gallaher Ltd'
SET @CategoryID = 377;
Declare @SELECTtQuery1 varchar(1000)
Declare @FromQuery1 varchar(1000)
Declare @GROUPBYQuery1 varchar(1000)
Declare @SQLQuery AS NVarchar(4000)
Declare @ParamDefinition AS NVarchar(2000)
SET @FromQuery1 = 'FROM [dbo].[iView_Data_TCG_Aggregated_Product]
WHERE PeriodName = '''+@Time+'''
AND Manufacturer = '''+@Manufacturer+''' AND CategoryId = '''+ @CategoryID+
print @FromQuery1
When i execute this query i am getting the conversion error. PLease find the error below.
----------------------
Conversion failed when converting the varchar value ' FROM [dbo].[iView_Data_TCG_Aggregated_Product]
WHERE PeriodName = '4 Weeks'
AND Manufacturer = 'Test1' AND CategoryId = '' to data type int.
----------------------------------------------
I am not getting categoryid value when i execute. It is giving error.
Please suggest me wht to do.
Thanks
Chandrakanth
|
|
|
|
|
A few things:
- Since you're concatenating things together, @CategoryID will have to be converted to a string. Otherwise, SQL Server is trying to perform addition and convert the first variables to an int.
Try changing the SET to this:
SET @FromQuery1 = 'FROM [dbo].[iView_Data_TCG_Aggregated_Product]
WHERE PeriodName = '''+@Time+'''
AND Manufacturer = '''+@Manufacturer+''' AND CategoryId = '''+ cast(@CategoryID as varchar(10))
- There was a dangling plus sign I removed from what you posted just to get it to execute
- If this is going to be a valid query, your'e missing the SELECT clause with the columns to select.
Scott
|
|
|
|
|
Hi Scott,Thanks and lot. I got it. It is working fine from my end .
Thanks
Chandra
|
|
|
|
|
My boss just tells me that I could have a table in the database that receives about 3x10^9 records a year (3 000 000 000). I could split this table in roughly four equal parts, but it still remains a lot. The user can choose to view just a day a year or even the entire history. (eg. the last 2 years wich still means 1 500 000 000 records)
I generally held a rule for myself to keep a production database around maximum 100 000 to 1 000 000 records total and put the rest in history, but this doesn't seem to be an option here and I'm very worried about the performance.
Has anyone got any experience with this size of database and how to solve the issue of getting the data quickly?
(Additonal info. The data are datapoints that need to be plotted on a graph in a certain time period. Performance is one of the top priorities.)
I see two major problems:
1. Keep table size reasonable, so how to split the data into different tables easily?
2. Getting the data to the user. (eg create average graph on the server, because a graph with more then, let's say 1000-2000 points is not useful on the screen anyway.)
Many thanks for any advice in advance.
(PS: Only thing I could find on google is splitting into smaller tables, but that doesn't immediately solve the problem of sending data to the user)
V.
|
|
|
|
|
Look into partitioning, probably by month.
Your users will NEVER look at 100k rows let alone 2 years worth. They may wish to analyse that volume but will now "view" them!
Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Actually, what I think will happen is that the "view" on large amounts will be averaged. for example one point per month or week.
I do realize I'll need to talk to the boss. 
V.
|
|
|
|
|
Using SQL Server to deal with large volumes of data takes some serious planning, if you need to get into this area I would engage a professional, while I consider myself a good SQL developer I would not undertake such a project without professional support.
We had such a requirement and ended up using Oracle, it is faster and more robust but you HAVE to have a full time DBA.
Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Mycroft Holmes wrote: Look into partitioning, probably by month.
Good call. 
"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
nils illegitimus carborundum
me, me, me
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
